| /packages/modules/NeuralNetworks/common/cpu_operations/ |
| D | QuantizedLSTMTest.cpp | 60 const uint32_t numBatches = inputOperandTypeParams[0].shape[0]; in QuantizedLSTMOpModel() local
67 OperandType cellStateOutOperandType(Type::TENSOR_QUANT16_SYMM, {numBatches, outputSize}, in QuantizedLSTMOpModel()
70 OperandType outputOperandType(Type::TENSOR_QUANT8_ASYMM, {numBatches, outputSize}, in QuantizedLSTMOpModel()
83 cellStateOut_.resize(numBatches * outputSize, 0); in QuantizedLSTMOpModel()
84 output_.resize(numBatches * outputSize, 0); in QuantizedLSTMOpModel()
240 const int numBatches = input.size(); in VerifyGoldens() local
241 EXPECT_GT(numBatches, 0); in VerifyGoldens()
248 for (int b = 0; b < numBatches; ++b) { in VerifyGoldens()
258 for (int b = 0; b < numBatches; ++b) { in VerifyGoldens()
272 const int numBatches = 2; in TEST_F() local
[all …]
|
| D | TransposeConv2D.cpp | 106 uint32_t numBatches = getSizeOfDimension(inputShape, 0); \
136 for (uint32_t b = 0; b < numBatches; b++) { in transposeConvNhwc()
166 const uint32_t outerSize = numBatches * outputHeight * outputWidth; in transposeConvNhwc()
222 for (uint32_t b = 0; b < numBatches; b++) { in transposeConvNhwc()
257 const uint32_t outerSize = numBatches * outputHeight * outputWidth; in transposeConvNhwc()
364 for (uint32_t b = 0; b < numBatches; b++) { in transposeConvQuant8PerChannelNhwc()
398 const uint32_t outerSize = numBatches * outputHeight * outputWidth; in transposeConvQuant8PerChannelNhwc()
|
| D | GenerateProposals.cpp | 71 uint32_t numBatches = getSizeOfDimension(imageInfoDataShape, 0); in bboxTransformFloat32() local
83 NN_RET_CHECK_LT(batchIndex, numBatches); in bboxTransformFloat32()
213 uint32_t numBatches = getSizeOfDimension(imageInfoShape, 0); in prepare() local
215 NN_RET_CHECK_GT(numBatches, 0u); in prepare()
855 uint32_t numBatches = getSizeOfDimension(scoresShape, 0); in generateProposalsNhwcFloat32Compute() local
898 for (uint32_t b = 0; b < numBatches; b++) { in generateProposalsNhwcFloat32Compute()
1128 uint32_t numBatches = getSizeOfDimension(scoreShape, 0); in prepare() local
1133 NN_RET_CHECK_EQ(getSizeOfDimension(bboxDeltasShape, 0), numBatches); in prepare()
1137 NN_RET_CHECK_EQ(getSizeOfDimension(imageInfoDataShape, 0), numBatches); in prepare()
1262 uint32_t numBatches = getSizeOfDimension(scoreShape, 0); in detectionPostprocessFloat32() local
[all …]
|
| D | RoiAlign.cpp | 58 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in roiAlignNhwc() local
77 NN_RET_CHECK_LT(batchId, numBatches); in roiAlignNhwc()
185 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in roiAlignQuantNhwc() local
209 NN_RET_CHECK_LT(batchId, numBatches); in roiAlignQuantNhwc()
340 uint32_t numBatches = getSizeOfDimension(input, 0); in prepare() local
346 NN_RET_CHECK_GT(numBatches, 0u); in prepare()
|
| D | GroupedConv2D.cpp | 38 uint32_t numBatches = getSizeOfDimension(inputShape, 0); \
64 for (uint32_t b = 0; b < numBatches; b++) { in groupedConvFloat32()
136 for (uint32_t b = 0; b < numBatches; b++) { in groupedConvQuant8()
241 for (uint32_t b = 0; b < numBatches; b++) { in groupedConvQuant8PerChannel()
|
| D | InstanceNormalization.cpp | 42 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in instanceNormNhwc() local
46 for (uint32_t b = 0; b < numBatches; b++) { in instanceNormNhwc()
|
| D | RoiPooling.cpp | 52 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in roiPoolingNhwc() local
71 NN_RET_CHECK_LT(batchId, numBatches); in roiPoolingNhwc()
186 [[maybe_unused]] uint32_t numBatches = getSizeOfDimension(input, 0); in prepare() local
|
| D | QuantizedLSTM.cpp | 255 const uint32_t numBatches = SizeOfDimension(input, 0); in prepare() local
260 NN_RET_CHECK_EQ(SizeOfDimension(prevOutput, 0), numBatches); in prepare()
320 NN_CHECK_EQ(SizeOfDimension(prevCellState, 0), numBatches); in prepare()
|
| D | Conv2D.cpp | 359 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in convQuant8PerChannelNhwc() local
394 for (uint32_t b = 0; b < numBatches; b++) { in convQuant8PerChannelNhwc()
452 [[maybe_unused]] uint32_t numBatches = getSizeOfDimension(inputShape, 0); in convQuant8PerChannelNhwc() local
|
| D | DepthwiseConv2D.cpp | 285 uint32_t numBatches = getSizeOfDimension(inputShape, 0); in depthwiseConvQuant8PerChannelNhwc() local
321 for (uint32_t b = 0; b < numBatches; b++) { in depthwiseConvQuant8PerChannelNhwc()
|
| /packages/modules/NeuralNetworks/runtime/test/ |
| D | TestValidateOperations.cpp | 3528 const int numBatches = 2; in detectionPostprocessingOpTest() local
3533 uint32_t inputDims[3] = {numBatches, numAnchors, numClasses}; in detectionPostprocessingOpTest()
3535 uint32_t deltasDims[3] = {numBatches, numAnchors, lengthBoxEncoding}; in detectionPostprocessingOpTest()
3552 uint32_t outputScoreDims[2] = {numBatches, maxNumDetectionsValue}; in detectionPostprocessingOpTest()
3554 uint32_t boundingBoxesDims[3] = {numBatches, maxNumDetectionsValue, 4}; in detectionPostprocessingOpTest()
3558 uint32_t numValidDims[1] = {numBatches}; in detectionPostprocessingOpTest()
|
| /packages/modules/NeuralNetworks/tools/api/ |
| D | types.spec | 5035 * and shape [numBatches, inputSize] specifying the input to the LSTM
5107 * and shape [numBatches, outputSize] specifying the cell state from the
5120 * and shape [numBatches, outputSize] which contains a cell state from
|