ArmNN
 24.02
RefUnidirectionalSequenceLstmWorkload.cpp
Go to the documentation of this file.
1 //
2 // Copyright © 2021-2023 Arm Ltd and Contributors. All rights reserved.
3 // SPDX-License-Identifier: MIT
4 //
5 
6 #include "RefUnidirectionalSequenceLstmWorkload.hpp"
7 #include "Activation.hpp"
8 #include "Encoders.hpp"
9 #include "Decoders.hpp"
10 #include "Lstm.hpp"
11 #include "LstmUtils.hpp"
12 #include "RefWorkloadUtils.hpp"
13 
14 #include <armnnUtils/Permute.hpp>
15 
16 namespace armnn
17 {
18 
19 RefUnidirectionalSequenceLstmWorkload::RefUnidirectionalSequenceLstmWorkload(
20  const UnidirectionalSequenceLstmQueueDescriptor& descriptor,
21  const WorkloadInfo& info)
22  : RefBaseWorkload<UnidirectionalSequenceLstmQueueDescriptor>(descriptor, info)
23  , m_InputToInputWeightsTensor (AssignScopedTensorHandle(descriptor.m_InputToInputWeights))
24  , m_InputToForgetWeightsTensor (AssignScopedTensorHandle(descriptor.m_InputToForgetWeights))
25  , m_InputToCellWeightsTensor (AssignScopedTensorHandle(descriptor.m_InputToCellWeights))
26  , m_InputToOutputWeightsTensor (AssignScopedTensorHandle(descriptor.m_InputToOutputWeights))
27  , m_RecurrentToInputWeightsTensor (AssignScopedTensorHandle(descriptor.m_RecurrentToInputWeights))
28  , m_RecurrentToForgetWeightsTensor(AssignScopedTensorHandle(descriptor.m_RecurrentToForgetWeights))
29  , m_RecurrentToCellWeightsTensor (AssignScopedTensorHandle(descriptor.m_RecurrentToCellWeights))
30  , m_RecurrentToOutputWeightsTensor(AssignScopedTensorHandle(descriptor.m_RecurrentToOutputWeights))
31  , m_CellToInputWeightsTensor (AssignScopedTensorHandle(descriptor.m_CellToInputWeights))
32  , m_CellToForgetWeightsTensor (AssignScopedTensorHandle(descriptor.m_CellToForgetWeights))
33  , m_CellToOutputWeightsTensor (AssignScopedTensorHandle(descriptor.m_CellToOutputWeights))
34  , m_InputGateBiasTensor (AssignScopedTensorHandle(descriptor.m_InputGateBias))
35  , m_ForgetGateBiasTensor (AssignScopedTensorHandle(descriptor.m_ForgetGateBias))
36  , m_CellBiasTensor (AssignScopedTensorHandle(descriptor.m_CellBias))
37  , m_OutputGateBiasTensor (AssignScopedTensorHandle(descriptor.m_OutputGateBias))
38  , m_ProjectionWeightsTensor (AssignScopedTensorHandle(descriptor.m_ProjectionWeights))
39  , m_ProjectionBiasTensor (AssignScopedTensorHandle(descriptor.m_ProjectionBias))
40  , m_InputLayerNormWeights (AssignScopedTensorHandle(descriptor.m_InputLayerNormWeights))
41  , m_ForgetLayerNormWeights (AssignScopedTensorHandle(descriptor.m_ForgetLayerNormWeights))
42  , m_CellLayerNormWeights (AssignScopedTensorHandle(descriptor.m_CellLayerNormWeights))
43  , m_OutputLayerNormWeights (AssignScopedTensorHandle(descriptor.m_OutputLayerNormWeights))
44 {}
45 
46 void RefUnidirectionalSequenceLstmWorkload::Execute() const
47 {
48  Execute(m_Data.m_Inputs, m_Data.m_Outputs);
49 }
50 
51 void RefUnidirectionalSequenceLstmWorkload::ExecuteAsync(ExecutionData& executionData)
52 {
53  WorkingMemDescriptor* workingMemDescriptor = static_cast<WorkingMemDescriptor*>(executionData.m_Data);
54  Execute(workingMemDescriptor->m_Inputs, workingMemDescriptor->m_Outputs);
55 }
56 
57 void RefUnidirectionalSequenceLstmWorkload::Execute(std::vector<ITensorHandle*> inputs,
58  std::vector<ITensorHandle*> outputs) const
59 {
60  ARMNN_SCOPED_PROFILING_EVENT_REF_NAME_GUID("RefUnidirectionalSequenceLstmWorkload_Execute");
61 
62  TensorInfo inputInfo = GetTensorInfo(inputs[0]);
63  const TensorInfo& outputStateInfo = GetTensorInfo(inputs[1]);
64  const TensorInfo& cellStateInfo = GetTensorInfo(inputs[2]);
65  TensorInfo outputStateOutInfo = GetTensorInfo(outputs[0]);
66  TensorInfo cellStateOutInfo = GetTensorInfo(outputs[1]);
67  TensorInfo outputInfo = GetTensorInfo(outputs[2]);
68  TensorShape& inputShape = inputInfo.GetShape();
69  TensorShape& outputShape= outputInfo.GetShape();
70  auto inputTensor = reinterpret_cast<float*>(inputs[0]->Map());
71 
72  if (!m_Data.m_Parameters.m_TimeMajor)
73  {
74  // Permute to time major
75  const PermutationVector& mappings = {1U, 0U, 2U};
76  std::vector<float> inputValue(inputTensor, inputTensor + inputInfo.GetNumElements());
77  inputShape = armnnUtils::Permuted(inputInfo.GetShape(), mappings);
78  inputInfo.SetShape(inputShape);
79  armnnUtils::Permute(inputShape, mappings, inputValue.data(), inputTensor, sizeof(float));
80 
81  outputShape = armnnUtils::Permuted(outputInfo.GetShape(), mappings);
82  outputInfo.SetShape(outputShape);
83  }
84  // As it is permuted to time major, maxTime is inputShape[0].
85  unsigned int maxTime = inputShape[0];
86  unsigned int batchSize = inputShape[1];
87  unsigned int outputSize = outputShape[2];
88  unsigned int inputSize = inputShape[2];
89 
90  TensorInfo scratchInfo = outputInfo;
91  scratchInfo.SetShape({batchSize, cellStateInfo.GetShape()[1]});
92 
93  std::vector<float> inputGateScratchBuffer;
94  std::vector<float> cellScratchBuffer(scratchInfo.GetNumElements(), 0.);
95  std::vector<float> forgetGateScratchBuffer(scratchInfo.GetNumElements(), 0.);
96  std::vector<float> outputGateScratchBuffer(scratchInfo.GetNumElements(), 0.);
97 
98  std::vector<float> outputStateOutBuffer(outputStateInfo.GetNumElements(), 0.);
99  std::vector<float> cellStateOutBuffer(cellStateInfo.GetNumElements(), 0.);
100 
101  void* outputStateOutData = outputStateOutBuffer.data();
102  void* cellStateOutData = cellStateOutBuffer.data();
103 
104  std::unique_ptr<Encoder<float>> inputGateScratch;
105  std::unique_ptr<Encoder<float>> cellScratch = MakeEncoder<float>(scratchInfo, cellScratchBuffer.data());
106  std::unique_ptr<Encoder<float>> forgetGateScratch = MakeEncoder<float>(scratchInfo, forgetGateScratchBuffer.data());
107  std::unique_ptr<Encoder<float>> outputGateScratch = MakeEncoder<float>(scratchInfo, outputGateScratchBuffer.data());
108 
109  std::unique_ptr<Decoder<float>> inputGateScratchDecoder;
110  std::unique_ptr<Decoder<float>> cellScratchDecoder = MakeDecoder<float>(scratchInfo, cellScratchBuffer.data());
111  std::unique_ptr<Decoder<float>> forgetGateScratchDecoder = MakeDecoder<float>(scratchInfo,
112  forgetGateScratchBuffer.data());
113  std::unique_ptr<Decoder<float>> outputGateScratchDecoder = MakeDecoder<float>(scratchInfo,
114  outputGateScratchBuffer.data());
115 
116  const bool useCifg = m_Data.m_Parameters.m_CifgEnabled;
117  const bool usePeephole = m_Data.m_Parameters.m_PeepholeEnabled;
118  const bool useLayerNorm = m_Data.m_Parameters.m_LayerNormEnabled;
119 
120  if (!useCifg)
121  {
122  inputGateScratchBuffer.resize(scratchInfo.GetNumElements(), 0.);
123  inputGateScratch = MakeEncoder<float>(scratchInfo, inputGateScratchBuffer.data());
124  inputGateScratchDecoder = MakeDecoder<float>(scratchInfo, inputGateScratchBuffer.data());
125  }
126 
127  std::unique_ptr<Encoder<float>> outputStateOut = MakeEncoder<float>(outputStateInfo, outputStateOutData);
128  std::unique_ptr<Encoder<float>> cellStateOut = MakeEncoder<float>(cellStateInfo, cellStateOutData);
129  std::unique_ptr<Decoder<float>> cellStateOutDecoder = MakeDecoder<float>(cellStateInfo, cellStateOutData);
130 
131  TensorInfo lstmInputInfo = inputInfo;
132  TensorShape batchInputShape = TensorShape({batchSize, inputSize});
133  lstmInputInfo.SetShape(batchInputShape);
134 
135  TensorInfo lstmOutputInfo = outputInfo;
136  lstmOutputInfo.SetShape({batchSize, outputSize});
137 
138  const TensorShape& inputToOutputWeightsShape = m_InputToOutputWeightsTensor->GetShape();
139  const TensorShape& recurrentToOutputWeightsShape = m_RecurrentToOutputWeightsTensor->GetShape();
140  unsigned int nOutput = recurrentToOutputWeightsShape[1];
141  auto outputStateInData = inputs[1]->Map();
142  std::unique_ptr<Decoder<float>> outputStateIn = MakeDecoder<float>(outputStateInfo, outputStateInData);
143 
144  auto cellStateInData = inputs[2]->Map();
145  std::unique_ptr<Decoder<float>> cellStateIn = MakeDecoder<float>(cellStateInfo, cellStateInData);
146 
147  auto currentInputData = reinterpret_cast<float*>(inputs[0]->Map());
148  std::unique_ptr<Decoder<float>> inputData = MakeDecoder<float>(lstmInputInfo, currentInputData);
149  auto currentOutputData = reinterpret_cast<float*>(outputs[2]->Map());
150  std::unique_ptr<Encoder<float>> output = MakeEncoder<float>(lstmOutputInfo, currentOutputData);
151  std::unique_ptr<Decoder<float>> outputDecoder = MakeDecoder<float>(lstmOutputInfo, currentOutputData);
152 
153  std::unique_ptr<Decoder<float>> inputToInputWeightsTensor;
154  std::unique_ptr<Decoder<float>> inputToForgetWeightsTensor = MakeDecoder<float>(
155  m_InputToForgetWeightsTensor->GetTensorInfo(), m_InputToForgetWeightsTensor->GetConstTensor<void>());
156  std::unique_ptr<Decoder<float>> inputToCellWeightsTensor = MakeDecoder<float>(
157  m_InputToCellWeightsTensor->GetTensorInfo(), m_InputToCellWeightsTensor->GetConstTensor<void>());
158  std::unique_ptr<Decoder<float>> inputToOutputWeightsTensor = MakeDecoder<float>(
159  m_InputToOutputWeightsTensor->GetTensorInfo(), m_InputToOutputWeightsTensor->GetConstTensor<void>());
160 
161  std::unique_ptr<Decoder<float>> recurrentToInputWeightsTensor;
162  std::unique_ptr<Decoder<float>> recurrentToForgetWeightsTensor = MakeDecoder<float>(
163  m_RecurrentToForgetWeightsTensor->GetTensorInfo(), m_RecurrentToForgetWeightsTensor->GetConstTensor<void>());
164  std::unique_ptr<Decoder<float>> recurrentToCellWeightsTensor = MakeDecoder<float>(
165  m_RecurrentToCellWeightsTensor->GetTensorInfo(), m_RecurrentToCellWeightsTensor->GetConstTensor<void>());
166  std::unique_ptr<Decoder<float>> recurrentToOutputWeightsTensor = MakeDecoder<float>(
167  m_RecurrentToOutputWeightsTensor->GetTensorInfo(), m_RecurrentToOutputWeightsTensor->GetConstTensor<void>());
168 
169  std::unique_ptr<Decoder<float>> inputGateBiasTensor;
170  std::unique_ptr<Decoder<float>> forgetGateBiasTensor = MakeDecoder<float>(
171  m_ForgetGateBiasTensor->GetTensorInfo(), m_ForgetGateBiasTensor->GetConstTensor<void>());
172  std::unique_ptr<Decoder<float>> cellBiasTensor = MakeDecoder<float>(
173  m_CellBiasTensor->GetTensorInfo(), m_CellBiasTensor->GetConstTensor<void>());
174  std::unique_ptr<Decoder<float>> outputGateBiasTensor = MakeDecoder<float>(
175  m_OutputGateBiasTensor->GetTensorInfo(), m_OutputGateBiasTensor->GetConstTensor<void>());
176 
177  std::unique_ptr<Decoder<float>> cellToInputWeightsTensor;
178  std::unique_ptr<Decoder<float>> cellToForgetWeightsTensor;
179  std::unique_ptr<Decoder<float>> cellToOutputWeightsTensor;
180 
181  std::unique_ptr<Decoder<float>> projectionWeightsTensor;
182  std::unique_ptr<Decoder<float>> projectionBiasTensor;
183 
184  std::unique_ptr<Decoder<float>> inputLayerNormWeights;
185  std::unique_ptr<Decoder<float>> forgetLayerNormWeights;
186  std::unique_ptr<Decoder<float>> cellLayerNormWeights;
187  std::unique_ptr<Decoder<float>> outputLayerNormWeights;
188 
189  if (useLayerNorm)
190  {
191  if (!useCifg)
192  {
193  inputLayerNormWeights = MakeDecoder<float>(
194  m_InputLayerNormWeights->GetTensorInfo(), m_InputLayerNormWeights->GetConstTensor<void>());
195  }
196  forgetLayerNormWeights = MakeDecoder<float>(
197  m_ForgetLayerNormWeights->GetTensorInfo(), m_ForgetLayerNormWeights->GetConstTensor<void>());
198  cellLayerNormWeights = MakeDecoder<float>(
199  m_CellLayerNormWeights->GetTensorInfo(), m_CellLayerNormWeights->GetConstTensor<void>());
200  outputLayerNormWeights = MakeDecoder<float>(
201  m_OutputLayerNormWeights->GetTensorInfo(), m_OutputLayerNormWeights->GetConstTensor<void>());
202  }
203 
204  if (!useCifg)
205  {
206  inputToInputWeightsTensor = MakeDecoder<float>(
207  m_InputToInputWeightsTensor->GetTensorInfo(), m_InputToInputWeightsTensor->GetConstTensor<void>());
208  inputGateBiasTensor = MakeDecoder<float>(
209  m_InputGateBiasTensor->GetTensorInfo(), m_InputGateBiasTensor->GetConstTensor<void>());
210  recurrentToInputWeightsTensor = MakeDecoder<float>(
211  m_RecurrentToInputWeightsTensor->GetTensorInfo(), m_RecurrentToInputWeightsTensor->GetConstTensor<void>());
212  }
213 
214  if (usePeephole)
215  {
216  cellToForgetWeightsTensor = MakeDecoder<float>(
217  m_CellToForgetWeightsTensor->GetTensorInfo(), m_CellToForgetWeightsTensor->GetConstTensor<void>());
218  cellToOutputWeightsTensor = MakeDecoder<float>(
219  m_CellToOutputWeightsTensor->GetTensorInfo(), m_CellToOutputWeightsTensor->GetConstTensor<void>());
220  }
221 
222  if (!useCifg && usePeephole)
223  {
224  cellToInputWeightsTensor = MakeDecoder<float>(
225  m_CellToInputWeightsTensor->GetTensorInfo(), m_CellToInputWeightsTensor->GetConstTensor<void>());
226  }
227 
228  if (m_Data.m_Parameters.m_ProjectionEnabled)
229  {
230  projectionWeightsTensor = MakeDecoder<float>(
231  m_ProjectionWeightsTensor->GetTensorInfo(), m_ProjectionWeightsTensor->GetConstTensor<void>());
232  if (m_ProjectionBiasTensor)
233  {
234  projectionBiasTensor = MakeDecoder<float>(
235  m_ProjectionBiasTensor->GetTensorInfo(), m_ProjectionBiasTensor->GetConstTensor<void>());
236  }
237  }
238 
239  unsigned int batchInputSize = batchSize * inputSize;
240  unsigned int batchOutputSize = batchSize * nOutput;
241 
242  for (unsigned int t = 0; t < maxTime; ++t)
243  {
244  LstmImpl(m_Data.m_Parameters,
245  lstmInputInfo,
246  lstmOutputInfo,
247  inputToOutputWeightsShape,
248  recurrentToOutputWeightsShape,
249  inputData,
250  outputStateIn,
251  cellStateIn,
252  outputStateOut,
253  cellStateOut,
254  output,
255  cellStateOutDecoder,
256  outputDecoder,
257  inputToInputWeightsTensor,
258  inputToForgetWeightsTensor,
259  inputToCellWeightsTensor,
260  inputToOutputWeightsTensor,
261  recurrentToInputWeightsTensor,
262  recurrentToForgetWeightsTensor,
263  recurrentToCellWeightsTensor,
264  recurrentToOutputWeightsTensor,
265  cellToInputWeightsTensor,
266  cellToForgetWeightsTensor,
267  cellToOutputWeightsTensor,
268  inputGateBiasTensor,
269  forgetGateBiasTensor,
270  cellBiasTensor,
271  outputGateBiasTensor,
272  projectionWeightsTensor,
273  projectionBiasTensor,
274  inputLayerNormWeights,
275  forgetLayerNormWeights,
276  cellLayerNormWeights,
277  outputLayerNormWeights,
278  inputGateScratch,
279  cellScratch,
280  forgetGateScratch,
281  outputGateScratch,
282  inputGateScratchDecoder,
283  cellScratchDecoder,
284  forgetGateScratchDecoder,
285  outputGateScratchDecoder,
286  m_LayerNormEpsilon);
287 
288  currentInputData += batchInputSize;
289  inputData = MakeDecoder<float>(lstmInputInfo, currentInputData);
290  currentOutputData += batchOutputSize;
291  output = MakeEncoder<float>(lstmOutputInfo, currentOutputData);
292  outputDecoder = MakeDecoder<float>(lstmOutputInfo, currentOutputData);
293 
294  // Assign output state out to the next output state in
295  outputStateIn = MakeDecoder<float>(outputStateInfo, outputStateOutData);
296 
297  // Assign cell state out to the next cell state in
298  cellStateIn = MakeDecoder<float>(cellStateInfo, cellStateOutData);
299  }
300 
301  if (!m_Data.m_Parameters.m_TimeMajor)
302  {
303  // Permute Output back to batch major
304  const PermutationVector& mappings = {1U, 0U, 2U};
305  auto outputData = reinterpret_cast<float*>(outputs[2]->Map());
306  std::vector<float> outputValue(outputData, outputData + outputInfo.GetNumElements());
307  outputShape = armnnUtils::Permuted(outputInfo.GetShape(), mappings);
308  outputInfo.SetShape(outputShape);
309  armnnUtils::Permute(outputShape, mappings, outputValue.data(), outputData, sizeof(float));
310  }
311 }
312 
313 } //namespace armnn
armnn::TensorInfo::GetNumElements
unsigned int GetNumElements() const
Definition: Tensor.hpp:198
armnn::LstmDescriptor::m_TimeMajor
bool m_TimeMajor
Enable/disable time major.
Definition: Descriptors.hpp:1154
Lstm.hpp
armnn::experimental::ExecutionData::m_Data
void * m_Data
Definition: ExecutionData.hpp:16
armnn::TensorInfo
Definition: Tensor.hpp:152
AssignScopedTensorHandle
std::unique_ptr< armnn::ScopedTensorHandle > AssignScopedTensorHandle(const armnn::ConstTensorHandle *ptr)
Definition: LstmUtils.cpp:299
armnn::RefUnidirectionalSequenceLstmWorkload::ExecuteAsync
void ExecuteAsync(ExecutionData &executionData) override
Definition: RefUnidirectionalSequenceLstmWorkload.cpp:51
ARMNN_SCOPED_PROFILING_EVENT_REF_NAME_GUID
#define ARMNN_SCOPED_PROFILING_EVENT_REF_NAME_GUID(label)
Creates a profiling event that uses GetGuid() and GetName() from the calling class.
Definition: RefWorkloadUtils.hpp:22
armnnUtils::Permute
void Permute(const armnn::TensorShape &dstShape, const armnn::PermutationVector &mappings, const void *src, void *dst, size_t dataTypeSize)
Definition: Permute.cpp:164
armnnUtils::Permuted
armnn::TensorShape Permuted(const armnn::TensorShape &srcShape, const armnn::PermutationVector &mappings)
Definition: Permute.cpp:125
armnn::LstmDescriptor::m_PeepholeEnabled
bool m_PeepholeEnabled
Enable/disable peephole.
Definition: Descriptors.hpp:1148
armnn::TensorShape
Definition: Tensor.hpp:20
armnn::LstmImpl
void LstmImpl(const LstmDescriptor &descriptor, const TensorInfo &inputInfo, const TensorInfo &outputInfo, const TensorShape &inputToOutputWeightsShape, const TensorShape &recurrentToOutputWeightsShape, std::unique_ptr< Decoder< float >> &inputData, std::unique_ptr< Decoder< float >> &outputStateIn, std::unique_ptr< Decoder< float >> &cellStateIn, std::unique_ptr< Encoder< float >> &outputStateOut, std::unique_ptr< Encoder< float >> &cellStateOut, std::unique_ptr< Encoder< float >> &output, std::unique_ptr< Decoder< float >> &cellStateOutDecoder, std::unique_ptr< Decoder< float >> &outputDecoder, std::unique_ptr< Decoder< float >> &inputToInputWeightsTensor, std::unique_ptr< Decoder< float >> &inputToForgetWeightsTensor, std::unique_ptr< Decoder< float >> &inputToCellWeightsTensor, std::unique_ptr< Decoder< float >> &inputToOutputWeightsTensor, std::unique_ptr< Decoder< float >> &recurrentToInputWeightsTensor, std::unique_ptr< Decoder< float >> &recurrentToForgetWeightsTensor, std::unique_ptr< Decoder< float >> &recurrentToCellWeightsTensor, std::unique_ptr< Decoder< float >> &recurrentToOutputWeightsTensor, std::unique_ptr< Decoder< float >> &cellToInputWeightsTensor, std::unique_ptr< Decoder< float >> &cellToForgetWeightsTensor, std::unique_ptr< Decoder< float >> &cellToOutputWeightsTensor, std::unique_ptr< Decoder< float >> &inputGateBiasTensor, std::unique_ptr< Decoder< float >> &forgetGateBiasTensor, std::unique_ptr< Decoder< float >> &cellBiasTensor, std::unique_ptr< Decoder< float >> &outputGateBiasTensor, std::unique_ptr< Decoder< float >> &projectionWeightsTensor, std::unique_ptr< Decoder< float >> &projectionBiasTensor, std::unique_ptr< Decoder< float >> &inputLayerNormWeights, std::unique_ptr< Decoder< float >> &forgetLayerNormWeights, std::unique_ptr< Decoder< float >> &cellLayerNormWeights, std::unique_ptr< Decoder< float >> &outputLayerNormWeights, std::unique_ptr< Encoder< float >> &inputGateScratch, std::unique_ptr< Encoder< float >> &cellScratch, std::unique_ptr< Encoder< float >> &forgetGateScratch, std::unique_ptr< Encoder< float >> &outputGateScratch, std::unique_ptr< Decoder< float >> &inputGateScratchDecoder, std::unique_ptr< Decoder< float >> &cellScratchDecoder, std::unique_ptr< Decoder< float >> &forgetGateScratchDecoder, std::unique_ptr< Decoder< float >> &outputGateScratchDecoder, float layerNormEpsilon)
Definition: Lstm.cpp:13
armnn::QueueDescriptorWithParameters::m_Parameters
LayerDescriptor m_Parameters
Definition: WorkloadData.hpp:66
armnn::WorkloadInfo
Contains information about TensorInfos of a layer.
Definition: WorkloadInfo.hpp:16
armnn::RefUnidirectionalSequenceLstmWorkload::RefUnidirectionalSequenceLstmWorkload
RefUnidirectionalSequenceLstmWorkload(const UnidirectionalSequenceLstmQueueDescriptor &descriptor, const WorkloadInfo &info)
Definition: RefUnidirectionalSequenceLstmWorkload.cpp:19
armnn::GetTensorInfo
const TensorInfo & GetTensorInfo(const ITensorHandle *tensorHandle)
float32 helpers
Definition: RefWorkloadUtils.hpp:33
Activation.hpp
armnn::PermutationVector
Definition: Types.hpp:314
Permute.hpp
armnn::BoostLogSeverityMapping::info
@ info
armnn::QueueDescriptor::m_Outputs
std::vector< ITensorHandle * > m_Outputs
Definition: WorkloadData.hpp:27
armnn::RefUnidirectionalSequenceLstmWorkload::Execute
void Execute() const override
Definition: RefUnidirectionalSequenceLstmWorkload.cpp:46
RefWorkloadUtils.hpp
armnn::LstmDescriptor::m_CifgEnabled
bool m_CifgEnabled
Enable/disable cifg (coupled input & forget gate).
Definition: Descriptors.hpp:1146
armnn::BaseWorkload< UnidirectionalSequenceLstmQueueDescriptor >::m_Data
UnidirectionalSequenceLstmQueueDescriptor m_Data
Definition: Workload.hpp:89
armnn::TensorInfo::GetShape
const TensorShape & GetShape() const
Definition: Tensor.hpp:193
armnn::LstmDescriptor::m_LayerNormEnabled
bool m_LayerNormEnabled
Enable/disable layer normalization.
Definition: Descriptors.hpp:1152
Decoders.hpp
armnn::experimental::WorkingMemDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkingMemDescriptor.hpp:20
armnn::TensorInfo::SetShape
void SetShape(const TensorShape &newShape)
Definition: Tensor.hpp:195
armnn
Copyright (c) 2021 ARM Limited and Contributors.
Definition: 01_00_quick_start.dox:6
armnn::experimental::WorkingMemDescriptor
Definition: WorkingMemDescriptor.hpp:18
armnn::LstmDescriptor::m_ProjectionEnabled
bool m_ProjectionEnabled
Enable/disable the projection layer.
Definition: Descriptors.hpp:1150
armnn::UnidirectionalSequenceLstmQueueDescriptor
Definition: WorkloadData.hpp:696
LstmUtils.hpp
Encoders.hpp
armnn::RefBaseWorkload
Definition: RefBaseWorkload.hpp:13
RefUnidirectionalSequenceLstmWorkload.hpp
armnn::experimental::WorkingMemDescriptor::m_Outputs
std::vector< ITensorHandle * > m_Outputs
Definition: WorkingMemDescriptor.hpp:21
armnn::QueueDescriptor::m_Inputs
std::vector< ITensorHandle * > m_Inputs
Definition: WorkloadData.hpp:26
armnn::experimental::ExecutionData
Definition: ExecutionData.hpp:14