Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / android-nn-driver / 990b5cf0fa562448f0590cf2650aa84cdc63f151 / . / ConversionUtils_1_2.hpp
blob: 779d88f4821b3803818053538c64f8cb8a2e528f [file] [log] [blame]
//
// Copyright © 2020 Arm Ltd and Contributors. All rights reserved.
// SPDX-License-Identifier: MIT
//
#pragma once
#include "Utils.hpp"
#include "ConversionUtils.hpp"
#include <armnn/utility/NumericCast.hpp>
#include <armnnUtils/TensorUtils.hpp>
#include <half/half.hpp>
using Half = half_float::half;
namespace armnn_driver
{
using namespace armnn;
using namespace android::nn;
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool IsQSymmDequantizeForWeights(const HalOperation& operation, const HalModel& model)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperationType = typename HalPolicy::OperationType;
const HalOperand* operand = GetInputOperand<HalPolicy>(operation, 0, model);
if (!operand)
{
return false;
}
if(!IsQSymm8(*operand))
{
// Only QSymm8 weights are dequantized on the fly by the driver
return false;
}
if (!IsOperandConstant<HalPolicy>(*operand))
{
// Non-const input is not accepted for weights
return false;
}
// Iterate through all the operations and find the operation feeding from the Dequantize output
const size_t outputIndex = operation.outputs[0];
for (uint32_t operationIdx = 0; operationIdx < getMainModel(model).operations.size(); ++operationIdx)
{
const auto& operationIt = getMainModel(model).operations[operationIdx];
switch (operationIt.type)
{
case HalOperationType::FULLY_CONNECTED:
if (outputIndex == operationIt.inputs[1]) // Weights are bound to slot 1
{
// If the output is going into the FC weights return true
return true;
}
break;
case HalOperationType::LSTM:
for (size_t k = 0; k < operationIt.inputs.size(); ++k)
{
if (outputIndex == operationIt.inputs[k])
{
// If the output is going into the LSTM weights return true
return true;
}
}
break;
default:
break;
}
}
return false;
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool SetupAndTrackLayerOutputSlotAndOverrideTensorInfo(const HalOperation& operation,
uint32_t operationOutputIndex,
armnn::IConnectableLayer& layer,
uint32_t layerOutputIndex,
const HalModel& model,
ConversionData& data,
const armnn::TensorInfo tensor_info)
{
using HalOperand = typename HalPolicy::Operand;
const HalOperand* outputOperand = GetOutputOperand<HalPolicy>(operation, operationOutputIndex, model);
if ((outputOperand == nullptr) || (operationOutputIndex >= layer.GetNumOutputSlots()))
{
return false;
}
armnn::IOutputSlot& outputSlot = layer.GetOutputSlot(layerOutputIndex);
const uint32_t operandIndex = operation.outputs[operationOutputIndex];
data.m_OutputSlotForOperand[operandIndex] = &outputSlot;
outputSlot.SetTensorInfo(tensor_info);
return true;
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertComparison_1_2(const HalOperation& operation,
const HalModel& model,
ConversionData& data,
ComparisonOperation comparisonOperation)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertComparison()");
ALOGV("comparisonOperation = %s", GetComparisonOperationAsCString(comparisonOperation));
LayerInputHandle input0 = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
LayerInputHandle input1 = ConvertToLayerInputHandle<HalPolicy>(operation, 1, model, data);
if (!(input0.IsValid() && input1.IsValid()))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo0 = input0.GetTensorInfo();
const TensorInfo& inputInfo1 = input1.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
ComparisonDescriptor descriptor(comparisonOperation);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsComparisonSupported,
data.m_Backends,
isSupported,
inputInfo0,
inputInfo1,
outputInfo,
descriptor);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddComparisonLayer(descriptor);
assert(layer != nullptr);
bool isReshapeSupported = BroadcastTensor(input0, input1, layer, data);
if (!isReshapeSupported)
{
return false;
}
if(IsDynamicTensor(outputInfo))
{
input0.Connect(layer->GetInputSlot(0));
input1.Connect(layer->GetInputSlot(1));
}
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertConv2d_1_2(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertConv2d_1_2()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
Convolution2dDescriptor desc;
desc.m_DataLayout = DataLayout::NHWC;
// Determine whether padding is implicit or explicit
bool implicitPadding = operation.inputs.size() == 7 ||
(operation.inputs.size() >= 8 &&
GetInputOperand<HalPolicy>(operation, 7, model)->type == HalOperandType::BOOL);
if (implicitPadding)
{
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 7, model, data);
}
else if (operation.inputs.size() >= 10)
{
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 10, model, data);
}
const PermutationVector OHWIToOIHW = {0, 2, 3, 1};
// ArmNN does not currently support non-fixed weights or bias
// The NNAPI filter is always OHWI [depth_out, filter_height, filter_width, depth_in] but ArmNN expects the
// filter's height and width indices to match the input's height and width indices so we permute it to OIHW if
// the DataLayout is NCHW
const ConstTensorPin weightsPin = (desc.m_DataLayout == DataLayout::NCHW) ?
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1,
model, data, OHWIToOIHW) :
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1, model, data);
const ConstTensorPin biasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 2, model, data);
if (!weightsPin.IsValid())
{
return Fail("%s: Operation has invalid weights", __func__);
}
if (!biasPin.IsValid())
{
return Fail("%s: Operation has invalid biases", __func__);
}
ConstTensor weights = weightsPin.GetConstTensor();
ConstTensor bias = biasPin.GetConstTensor();
SanitizeBiasQuantizationScale(bias.GetInfo(), weights.GetInfo(), inputInfo);
ActivationFn activation;
if (implicitPadding)
{
android::nn::PaddingScheme paddingScheme;
if (!GetInputPaddingScheme<HalPolicy>(operation, 3, paddingScheme, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 6, activation, model, data) ||
!GetOptionalConvolutionDilationParams<HalPolicy>(operation, 8, desc, model, data))
{
return Fail("%s: Operation has invalid inputs (implicit padding)", __func__);
}
armnnUtils::DataLayoutIndexed dataLayoutIndexed(desc.m_DataLayout);
unsigned int widthIndex = dataLayoutIndexed.GetWidthIndex();
unsigned int heightIndex = dataLayoutIndexed.GetHeightIndex();
const uint32_t kernelX = weights.GetShape()[widthIndex];
const uint32_t kernelY = weights.GetShape()[heightIndex];
const uint32_t inputX = inputInfo.GetShape()[widthIndex];
const uint32_t inputY = inputInfo.GetShape()[heightIndex];
CalcPadding(inputX, kernelX, desc.m_StrideX, desc.m_DilationX, desc.m_PadLeft, desc.m_PadRight, paddingScheme);
CalcPadding(inputY, kernelY, desc.m_StrideY, desc.m_DilationY, desc.m_PadTop, desc.m_PadBottom, paddingScheme);
}
else if (operation.inputs.size() >= 10)
{
// explicit padding
if (!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::INT32, desc.m_PadLeft, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_PadRight, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_PadTop, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, desc.m_PadBottom, model, data) ||
!GetInputScalar<HalPolicy>(operation, 7, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 8, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 9, activation, model, data) ||
!GetOptionalConvolutionDilationParams<HalPolicy>(operation, 11, desc, model, data))
{
return Fail("%s: Operation has invalid inputs (explicit padding)", __func__);
}
}
else
{
return Fail("%s: Unsupported number of operation inputs", __func__);
}
desc.m_BiasEnabled = true;
Optional<TensorInfo> biases(bias.GetInfo());
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsConvolution2dSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
desc,
weights.GetInfo(),
biases);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* startLayer =
data.m_Network->AddConvolution2dLayer(desc, weights, Optional<ConstTensor>(bias));
if (!startLayer)
{
return Fail("%s: AddConvolution2dLayer failed", __func__);
}
input.Connect(startLayer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *startLayer, model,
data, nullptr, validateFunc, activation);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertDepthwiseConv2d_1_2(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertDepthwiseConv2d_1_2()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// ArmNN does not currently support non-fixed weights or bias
// Find the shape of the weights tensor. In AndroidNN this will be [ 1, H, W, I * M ]
const HalOperand* weightsOperand = GetInputOperand<HalPolicy>(operation, 1, model);
if (weightsOperand == nullptr)
{
return Fail("%s: Operand is invalid", __func__);
}
if ( weightsOperand->dimensions[0] != 1)
{
return Fail("%s: Invalid weights; for depthwise convolution, dimension 0 must be 1 but it is %i",
__func__, weightsOperand->dimensions[0] );
}
DepthwiseConvolution2dDescriptor desc;
desc.m_DataLayout = DataLayout::NHWC;
// Determine whether padding is implicit or explicit
bool implicitPadding = operation.inputs.size() == 8 ||
(operation.inputs.size() >= 9 &&
GetInputOperand<HalPolicy>(operation, 8, model)->type == HalOperandType::BOOL);
// Look ahead to find the optional DataLayout, if present
const uint32_t dataLayoutFlagIndex = implicitPadding ? 8 : 11;
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, dataLayoutFlagIndex, model, data);
armnnUtils::DataLayoutIndexed dataLayoutIndexed(desc.m_DataLayout);
unsigned int channelsIndex = dataLayoutIndexed.GetChannelsIndex();
unsigned int widthIndex = dataLayoutIndexed.GetWidthIndex();
unsigned int heightIndex = dataLayoutIndexed.GetHeightIndex();
// Reinterpret weight data as [ H, W, I, M ]
TensorShape weightsShape({ weightsOperand->dimensions[1],
weightsOperand->dimensions[2],
inputInfo.GetShape()[channelsIndex],
weightsOperand->dimensions[3] / inputInfo.GetShape()[channelsIndex] });
// Swizzle weight data [ H, W, I, M ] -> [ M, I, H, W ]
const PermutationVector HWIMToMIHW = { 2U, 3U, 1U, 0U };
const ConstTensorPin weightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
1,
model,
data,
HWIMToMIHW,
&weightsShape);
// Bias is a 1D tensor
const ConstTensorPin biasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 2, model, data);
if (!weightsPin.IsValid())
{
return Fail("%s: Operation has invalid weights", __func__);
}
if (!biasPin.IsValid())
{
return Fail("%s: Operation has invalid biases", __func__);
}
ConstTensor weights = weightsPin.GetConstTensor();
ConstTensor bias = biasPin.GetConstTensor();
SanitizeBiasQuantizationScale(bias.GetInfo(), weights.GetInfo(), inputInfo);
ActivationFn activation;
if (implicitPadding)
{
android::nn::PaddingScheme paddingScheme;
if (!GetInputPaddingScheme<HalPolicy>(operation, 3, paddingScheme, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 7, activation, model, data) ||
!GetOptionalConvolutionDilationParams<HalPolicy>(operation, 9, desc, model, data))
{
return Fail("%s: Operation has invalid inputs (implicit padding)", __func__);
}
const uint32_t kernelX = weights.GetShape()[3];
const uint32_t kernelY = weights.GetShape()[2];
const uint32_t inputX = inputInfo.GetShape()[widthIndex];
const uint32_t inputY = inputInfo.GetShape()[heightIndex];
CalcPadding(inputX, kernelX, desc.m_StrideX, desc.m_DilationX, desc.m_PadLeft, desc.m_PadRight, paddingScheme);
CalcPadding(inputY, kernelY, desc.m_StrideY, desc.m_DilationY, desc.m_PadTop, desc.m_PadBottom, paddingScheme);
}
else if (operation.inputs.size() >= 11)
{
// explicit padding
if (!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::INT32, desc.m_PadLeft, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_PadRight, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_PadTop, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, desc.m_PadBottom, model, data) ||
!GetInputScalar<HalPolicy>(operation, 7, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 8, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 10, activation, model, data) ||
!GetOptionalConvolutionDilationParams<HalPolicy>(operation, 12, desc, model, data))
{
return Fail("%s: Operation has invalid inputs (explicit padding)", __func__);
}
}
else
{
return Fail("%s: Unsupported number of operation inputs", __func__);
}
desc.m_BiasEnabled = true;
Optional<TensorInfo> biases(bias.GetInfo());
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsDepthwiseConvolutionSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
desc,
weights.GetInfo(),
biases);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* startLayer =
data.m_Network->AddDepthwiseConvolution2dLayer(desc, weights, Optional<ConstTensor>(bias));
if (!startLayer)
{
return Fail("%s: AddDepthwiseConvolution2dLayer failed", __func__);
}
input.Connect(startLayer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *startLayer, model,
data, nullptr, validateFunc, activation);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertDequantize_1_2(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
ALOGV("HalPolicy::ConvertDequantize()");
if (IsQSymmDequantizeForWeights<HalPolicy>(operation, model))
{
// NOTE: QSymm8 weights are dequantized internally by the driver,
// therefore this type of Dequantize is implicitly supported
return true;
}
return ::ConvertDequantize<HalPolicy>(operation, model, data);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertElementwiseUnary(const HalOperation& operation,
const HalModel& model,
ConversionData& data,
UnaryOperation unaryOperation)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertElementwiseUnary()");
ALOGV("unaryOperation = %s", GetUnaryOperationAsCString(unaryOperation));
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid input", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
ElementwiseUnaryDescriptor descriptor(unaryOperation);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsElementwiseUnarySupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
descriptor);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddElementwiseUnaryLayer(descriptor);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertExpandDims(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertExpandDims()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid input", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Operation has invalid output", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
int32_t axis;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::INT32, axis, model, data))
{
return Fail("%s: failed to get axis input value", __func__);
}
TensorShape targetShape;
try
{
targetShape = armnnUtils::ExpandDims(input.GetTensorInfo().GetShape(), axis);
}
catch (const std::exception& e)
{
return Fail("%s: %s", __func__, e.what());
}
if (targetShape != outputInfo.GetShape())
{
return Fail("%s: Shape of the output operand does not match the resolved expanded shape", __func__);
}
ReshapeDescriptor reshapeDescriptor;
reshapeDescriptor.m_TargetShape = targetShape;
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsReshapeSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
outputInfo,
reshapeDescriptor);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddReshapeLayer(reshapeDescriptor);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertGather(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertGather()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid input", __func__);
}
auto inputDimensions = input.GetTensorInfo().GetNumDimensions();
LayerInputHandle indices = ConvertToLayerInputHandle<HalPolicy>(operation, 2, model, data);
if (!indices.IsValid())
{
return Fail("%s: Operation has invalid indices", __func__);
}
auto indicesDimensions = indices.GetTensorInfo().GetNumDimensions();
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Operation has invalid output", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
auto outputDimensions = outputInfo.GetNumDimensions();
if (outputDimensions != inputDimensions + indicesDimensions - 1)
{
return Fail("%s: Operation has invalid output dimensions: %d. Output must be an (%d + %d - 1)-D tensor",
__func__, outputDimensions, inputDimensions, indicesDimensions);
}
int32_t axis;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::INT32, axis, model, data))
{
return Fail("%s: Operation has invalid or unsupported axis operand", __func__);
}
if (((axis < -inputDimensions) && (axis < 0)) || ((axis >= inputDimensions) && (axis > 0)))
{
return Fail("%s: Operation has invalid axis: %d. It is out of bounds [-%d, %d))", __func__, axis,
inputDimensions, inputDimensions);
}
GatherDescriptor desc;
desc.m_Axis = axis;
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsGatherSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
indices.GetTensorInfo(),
outputInfo,
desc);
};
if(!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddGatherLayer(desc);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
indices.Connect(layer->GetInputSlot(1));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertGroupedConv2d(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertGroupedConv2d()");
//
// Parse data
//
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
TensorInfo outputInfo = GetTensorInfoForOperand(*output);
// Look ahead to determine data layout
DataLayout dataLayout = DataLayout::NHWC;
if (operation.inputs.size() == 12)
{
dataLayout = OptionalDataLayout<HalPolicy>(operation, 11, model, data);
}
else
{
dataLayout = OptionalDataLayout<HalPolicy>(operation, 8, model, data);
}
// NOTE:
// NNAPI weights are always OHWI, i.e. [depth_out, filter_height, filter_width, depth_group],
// but Arm NN expects the filter's height and width indices to match the input's height and
// width indices so when the DataLayout is NCHW, we need to permute the weights to OIHW
const PermutationVector ohwiToOihw = { 0u, 2u, 3u, 1u };
const ConstTensorPin weightsPin = (dataLayout == DataLayout::NCHW) ?
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1,
model, data, ohwiToOihw) :
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1, model, data);
const ConstTensorPin biasesPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 2, model, data);
if (!weightsPin.IsValid() || !biasesPin.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
ConstTensor weights = weightsPin.GetConstTensor();
ConstTensor biases = biasesPin.GetConstTensor();
SanitizeBiasQuantizationScale(biases.GetInfo(), weights.GetInfo(), inputInfo);
const TensorShape& inputShape = inputInfo.GetShape();
const TensorShape& outputShape = outputInfo.GetShape();
const TensorShape& weightsShape = weights.GetShape();
const TensorShape& biasesShape = biases.GetShape();
armnnUtils::DataLayoutIndexed dataLayoutIndexed(dataLayout);
const unsigned int channelsIndex = dataLayoutIndexed.GetChannelsIndex();
const unsigned int heightIndex = dataLayoutIndexed.GetHeightIndex();
const unsigned int widthIndex = dataLayoutIndexed.GetWidthIndex();
Convolution2dDescriptor desc;
desc.m_DataLayout = dataLayout;
desc.m_BiasEnabled = true;
int numGroups;
ActivationFn activation;
if (operation.inputs.size() == 12)
{
if (!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::INT32, desc.m_PadLeft, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_PadRight, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_PadTop, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, desc.m_PadBottom, model, data) ||
!GetInputScalar<HalPolicy>(operation, 7, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 8, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputScalar<HalPolicy>(operation, 9, HalOperandType::INT32, numGroups, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 10, activation, model, data))
{
return Fail("%s: Operation has invalid inputs (explicit padding)", __func__);
}
}
else if (operation.inputs.size() == 9)
{
android::nn::PaddingScheme paddingScheme;
if (!GetInputPaddingScheme<HalPolicy>(operation, 3, paddingScheme, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, numGroups, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 7, activation, model, data))
{
return Fail("%s: Operation has invalid inputs (implicit padding)", __func__);
}
const uint32_t inputX = inputInfo.GetShape()[widthIndex];
const uint32_t inputY = inputInfo.GetShape()[heightIndex];
const uint32_t kernelX = weightsShape[widthIndex];
const uint32_t kernelY = weightsShape[heightIndex];
CalcPadding(inputX, kernelX, desc.m_StrideX, desc.m_PadLeft, desc.m_PadRight, paddingScheme);
CalcPadding(inputY, kernelY, desc.m_StrideY, desc.m_PadTop, desc.m_PadBottom, paddingScheme);
}
else
{
return Fail("%s: Unsupported number of operation inputs", __func__);
}
// Equivalent to outputShape[channelsIndex], but we can't know the outputShape in the case of dynamic tensors
const unsigned int outputChannels = weightsShape[0];
const unsigned int channelsPerGroup = weightsShape[channelsIndex];
const unsigned int channelMultiplier = outputChannels / numGroups;
//
// Validate all relevant inputs
//
if (numGroups <= 0)
{
return Fail("%s: Number of groups must be greater than 0. Got: %d", __func__, numGroups);
}
if (outputChannels % numGroups != 0u)
{
return Fail("%s: Output channels must be divisible by the number of groups", __func__);
}
//
// Set up Splitter layer
//
unsigned int splitterDimSizes[4] = { inputShape[0], inputShape[1], inputShape[2], inputShape[3] };
splitterDimSizes[channelsIndex] /= numGroups; // split in depth
TensorInfo splitterOutputInfo(4,
splitterDimSizes,
inputInfo.GetDataType(),
inputInfo.GetQuantizationScale(),
inputInfo.GetQuantizationOffset());
std::vector<std::reference_wrapper<TensorInfo>> splitterOutputInfos(numGroups, std::ref(splitterOutputInfo));
ViewsDescriptor splitterDesc(numGroups);
for (unsigned int group = 0u; group < numGroups; ++group)
{
splitterDesc.SetViewOriginCoord(group, channelsIndex, splitterDimSizes[channelsIndex] * group);
for (unsigned int dimIdx = 0u; dimIdx < 4u; dimIdx++)
{
splitterDesc.SetViewSize(group, dimIdx, splitterDimSizes[dimIdx]);
}
}
bool isSupported = false;
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsSplitterSupported,
data.m_Backends,
isSupported,
inputInfo,
splitterOutputInfos,
splitterDesc);
if (!isSupported)
{
return false;
}
IConnectableLayer* splitterLayer = data.m_Network->AddSplitterLayer(splitterDesc);
if (!splitterLayer)
{
return Fail("%s: Failed to add SplitterLayer", __func__);
}
input.Connect(splitterLayer->GetInputSlot(0));
for (unsigned int group = 0u; group < splitterLayer->GetNumOutputSlots(); ++group)
{
splitterLayer->GetOutputSlot(group).SetTensorInfo(splitterOutputInfo);
}
//
// Set up Convolution2d layers for each group
//
// Set up group tensor shapes
TensorShape groupInputShape(inputShape);
groupInputShape[channelsIndex] = channelsPerGroup;
TensorShape groupWeightsShape(weightsShape);
groupWeightsShape[0] /= channelMultiplier * numGroups;
TensorShape groupBiasesShape({ 1 });
// Set up group tensor infos
TensorInfo groupInputInfo(inputInfo);
groupInputInfo.SetShape(groupInputShape);
const TensorInfo& weightsInfo = weights.GetInfo();
TensorInfo groupWeightsInfo(weightsInfo);
groupWeightsInfo.SetShape(groupWeightsShape);
const TensorInfo& biasesInfo = biases.GetInfo();
TensorInfo groupBiasesInfo(biasesInfo);
groupBiasesInfo.SetShape(groupBiasesShape);
TensorInfo groupOutputInfo(outputInfo);
TensorShape groupOutputShape(outputShape);
const bool isDynamic = IsDynamicTensor(outputInfo);
if (!isDynamic)
{
groupOutputShape[channelsIndex] = 1;
}
groupOutputInfo.SetShape(groupOutputShape);
const unsigned int weightsDataTypeSize = GetDataTypeSize(groupWeightsInfo.GetDataType());
const unsigned int biasesDataTypeSize = GetDataTypeSize(groupBiasesInfo.GetDataType());
std::vector<IConnectableLayer*> convLayers(numGroups * channelMultiplier, nullptr);
for (unsigned int group = 0u; group < numGroups; ++group)
{
for (unsigned int m = 0u; m < channelMultiplier; ++m)
{
auto index = group * channelMultiplier + m;
const unsigned int weightsDataOffset = groupWeightsShape.GetNumElements() * index * weightsDataTypeSize;
const unsigned int biasesDataOffset = groupBiasesShape.GetNumElements() * index * biasesDataTypeSize;
if (weightsInfo.HasPerAxisQuantization())
{
// Extract per-axis quantization scales for group weights
const std::vector<float>& weightsQuantScales = weightsInfo.GetQuantizationScales();
groupWeightsInfo.SetQuantizationScales(
std::vector<float>(weightsQuantScales.begin() + index,
weightsQuantScales.begin() + index + groupWeightsShape[0]));
// Extract per-axis quantization scales for group biases
const std::vector<float>& biasesQuantScales = biasesInfo.GetQuantizationScales();
groupBiasesInfo.SetQuantizationScales(
std::vector<float>(biasesQuantScales.begin() + index,
biasesQuantScales.begin() + index + groupWeightsShape[0]));
}
// Extract weights and biases data for current group convolution
ConstTensor groupWeights(groupWeightsInfo,
static_cast<const void *>(reinterpret_cast<const char *>(weights.GetMemoryArea()) +
weightsDataOffset));
ConstTensor groupBiases(groupBiasesInfo,
static_cast<const void *>(reinterpret_cast<const char *>(biases.GetMemoryArea()) +
biasesDataOffset));
isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsConvolution2dSupported,
data.m_Backends,
isSupported,
groupInputInfo,
outputInfo,
desc,
groupWeightsInfo,
Optional<TensorInfo>(groupBiasesInfo));
};
if(!isDynamic)
{
validateFunc(groupOutputInfo, isSupported);
}
else
{
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* convLayer =
data.m_Network->AddConvolution2dLayer(desc, groupWeights, Optional<ConstTensor>(groupBiases));
if (!convLayer)
{
return Fail("%s: AddConvolution2dLayer failed", __func__);
}
splitterLayer->GetOutputSlot(group).Connect(convLayer->GetInputSlot(0));
convLayer->GetOutputSlot(0).SetTensorInfo(groupOutputInfo);
if(isDynamic)
{
convLayer->GetOutputSlot(0).IsTensorInfoSet();
validateFunc(convLayer->GetOutputSlot(0).GetTensorInfo(), isSupported);
outputInfo = convLayer->GetOutputSlot(0).GetTensorInfo();
if (!isSupported)
{
return false;
}
}
convLayers[index] = convLayer;
}
}
//
// Set up Concat layer
//
ConcatDescriptor concatDescriptor;
// Equivalent to outputShape[channelsIndex], but we can't know the outputShape in the case of dynamic tensors
concatDescriptor = ConcatDescriptor(weightsShape[0]);
for (unsigned int group = 0u; group < numGroups; ++group)
{
for (unsigned int m = 0u; m < channelMultiplier; ++m)
{
auto index = group * channelMultiplier + m;
concatDescriptor.SetViewOriginCoord(index, channelsIndex, index);
concatDescriptor.SetConcatAxis(channelsIndex);
}
}
isSupported = false;
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsConcatSupported,
data.m_Backends,
isSupported,
std::vector<const TensorInfo*>(numGroups * channelMultiplier, &groupOutputInfo),
outputInfo,
concatDescriptor);
if (!isSupported)
{
return false;
}
IConnectableLayer* concatLayer = data.m_Network->AddConcatLayer(concatDescriptor);
if (!concatLayer)
{
return Fail("%s: AddConcatLayer failed", __func__);
}
for (unsigned int group = 0u; group < numGroups; ++group)
{
for (unsigned int m = 0u; m < channelMultiplier; ++m)
{
auto index = group * channelMultiplier + m;
convLayers[index]->GetOutputSlot(0).Connect(concatLayer->GetInputSlot(index));
}
}
concatLayer->GetOutputSlot(0).SetTensorInfo(outputInfo);
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *concatLayer, model,
data, nullptr, nullptr, activation);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertInstanceNormalization(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertInstanceNormalization()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has an invalid input 0", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Operation has an invalid output", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// Determine data type of input tensor
HalOperandType inputType;
if (!GetOperandType<HalPolicy>(operation, 0, model, inputType))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
InstanceNormalizationDescriptor desc;
// Read gamma, beta & epsilon
if (inputType == HalOperandType::TENSOR_FLOAT16)
{
Half fp16Gamma;
Half fp16Beta;
Half fp16Epsilon;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT16, fp16Gamma, model, data) ||
!GetInputScalar<HalPolicy>(operation, 2, HalOperandType::FLOAT16, fp16Beta, model, data) ||
!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::FLOAT16, fp16Epsilon, model, data))
{
return Fail("%s: Operation has invalid inputs (FLOAT16)", __func__);
}
desc.m_Gamma = static_cast<float>(fp16Gamma);
desc.m_Beta = static_cast<float>(fp16Beta);
desc.m_Eps = static_cast<float>(fp16Epsilon);
}
else if (inputType == HalOperandType::TENSOR_FLOAT32)
{
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT32, desc.m_Gamma, model, data) ||
!GetInputScalar<HalPolicy>(operation, 2, HalOperandType::FLOAT32, desc.m_Beta, model, data) ||
!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::FLOAT32, desc.m_Eps, model, data))
{
return Fail("%s: Operation has invalid inputs (FLOAT32)", __func__);
}
}
else
{
return Fail("%s: Unsupported input tensor type: %d", __func__, inputType);
}
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 4, model, data);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsInstanceNormalizationSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
outputInfo,
desc);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddInstanceNormalizationLayer(desc);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertLogSoftmax(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertLogSoftmax()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Failed to read input 0", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Failed to read output", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// Determine data type of input tensor
HalOperandType inputType;
if (!GetOperandType<HalPolicy>(operation, 0, model, inputType))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
LogSoftmaxDescriptor descriptor;
// Read beta
if (inputType == HalOperandType::TENSOR_FLOAT16)
{
Half fp16Beta;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT16, fp16Beta, model, data))
{
return Fail("%s: Failed to read input 1 (FLOAT16)", __func__);
}
descriptor.m_Beta = static_cast<float>(fp16Beta);
}
else if (inputType == HalOperandType::TENSOR_FLOAT32)
{
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT32, descriptor.m_Beta, model, data))
{
return Fail("%s: Failed to read input 1 (FLOAT32)", __func__);
}
}
else
{
return Fail("%s: Unsupported input tensor type: %d", __func__, inputType);
}
// Read axis
if (!GetInputInt32<HalPolicy>(operation, 2, descriptor.m_Axis, model, data))
{
return Fail("%s: Failed to read input 2", __func__);
}
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsLogSoftmaxSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
outputInfo,
descriptor);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddLogSoftmaxLayer(descriptor);
if (!layer)
{
return Fail("%s: AddLogSoftmaxLayer() returned nullptr", __func__);
}
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertMaximum(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertMaximum()");
LayerInputHandle input0 = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
LayerInputHandle input1 = ConvertToLayerInputHandle<HalPolicy>(operation, 1, model, data);
if (!input0.IsValid() || !input1.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* outputOperand = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!outputOperand)
{
return Fail("%s: Could not read output", __func__);
}
const TensorInfo& outInfo = GetTensorInfoForOperand(*outputOperand);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsMaximumSupported,
data.m_Backends,
isSupported,
input0.GetTensorInfo(),
input1.GetTensorInfo(),
outInfo);
};
if(IsDynamicTensor(outInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddMaximumLayer();
assert(layer != nullptr);
bool isReshapeSupported = BroadcastTensor(input0, input1, layer, data);
if (!isReshapeSupported)
{
return false;
}
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertMinimum(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertMinimum()");
LayerInputHandle input0 = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
LayerInputHandle input1 = ConvertToLayerInputHandle<HalPolicy>(operation, 1, model, data);
if (!input0.IsValid() || !input1.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsMinimumSupported,
data.m_Backends,
isSupported,
input0.GetTensorInfo(),
input1.GetTensorInfo(),
outputInfo);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddMinimumLayer();
assert(layer != nullptr);
bool isReshapeSupported = BroadcastTensor(input0, input1, layer, data);
if (!isReshapeSupported)
{
return false;
}
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertPadV2(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertPadV2()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Could not read input 0", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
unsigned int rank = inputInfo.GetNumDimensions();
PadDescriptor descriptor;
if (!ConvertPaddings<HalPolicy>(operation, model, data, rank, descriptor))
{
return Fail("%s: Could not convert paddings", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// Determine type of padding value
HalOperandType operandType0;
HalOperandType operandType2;
if (!GetOperandType<HalPolicy>(operation, 0, model, operandType0) ||
!GetOperandType<HalPolicy>(operation, 2, model, operandType2))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
// Read value to use for padding
if (operandType0 == HalOperandType::TENSOR_FLOAT16 && operandType2 == HalOperandType::FLOAT16)
{
Half f16PadValue;
if (!GetInputScalar<HalPolicy>(operation, 2, operandType2, f16PadValue, model, data))
{
return Fail("%s: Could not read input 2 (FLOAT16)", __func__);
}
descriptor.m_PadValue = f16PadValue;
}
else if (operandType0 == HalOperandType::TENSOR_FLOAT32 && operandType2 == HalOperandType::FLOAT32)
{
if (!GetInputFloat32<HalPolicy>(operation, 2, descriptor.m_PadValue, model, data))
{
return Fail("%s: Could not read input 2 (FLOAT32)", __func__);
}
}
else if (operandType0 == HalOperandType::TENSOR_QUANT8_ASYMM && operandType2 == HalOperandType::INT32)
{
int32_t intPadValue = 0;
if (!GetInputInt32<HalPolicy>(operation, 2, intPadValue, model, data))
{
return Fail("%s: Could not read input 2 (INT32)", __func__);
}
descriptor.m_PadValue = intPadValue;
}
else
{
return Fail("%s: Operation has invalid inputs: type mismatch", __func__);
}
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsPadSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
descriptor);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddPadLayer(descriptor);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertPrelu(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertPrelu()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
LayerInputHandle alpha = ConvertToLayerInputHandle<HalPolicy>(operation, 1, model, data);
if (!input.IsValid() || !alpha.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& alphaInfo = alpha.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsPreluSupported,
data.m_Backends,
isSupported,
inputInfo,
alphaInfo,
outputInfo);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddPreluLayer();
if (!layer)
{
return Fail("%s: AddPreluLayer failed", __func__);
}
bool isReshapeSupported = BroadcastTensor(input, alpha, layer, data);
if (!isReshapeSupported)
{
return false;
}
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertQuantize(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertQuantize()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid input", __func__);
}
const HalOperand* const outputOperand = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!outputOperand)
{
return Fail("%s: Operation has invalid outputs", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*outputOperand);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsQuantizeSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
outputInfo);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddQuantizeLayer();
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertQuantized16BitLstm(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
ALOGV("HalPolicy::ConvertQuantized16BitLstm()");
//Inputs:
// 0: The input: A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape [numBatches, inputSize]
// specifying the input to the LSTM cell. Tensor is quantized with a fixed quantization range of -1, 127/128.
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Could not read input 0: input", __func__);
}
//13: The previous cell state: A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT16_SYMM and shape
// [numBatches, outputSize] specifying the cell state from the previous time step of the LSTM cell.
// It is quantized using a quantization range of -2^4, 2^4 * 32767/32768.
LayerInputHandle previousCellStateIn = ConvertToLayerInputHandle<HalPolicy>(operation, 13, model, data);
if (!previousCellStateIn.IsValid())
{
return Fail("%s: Could not read input 13: previousCellStateIn", __func__);
}
// 14: The previous output state: A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [numBathes, outputSize] specifying the output of the LSTM cell from previous time-step. Tensor
// is quantized with a fixed quantization range of -1, 127/128.
LayerInputHandle previousOutputIn = ConvertToLayerInputHandle<HalPolicy>(operation, 14, model, data);
if (!previousOutputIn.IsValid())
{
return Fail("%s: Could not read input 14: previousOutputIn", __func__);
}
// Get the input tensors:
// 1: The input-to-input weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, inputSize] specifying input-to-input part of weights for fully-connected layer inside the
// LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin inputToInputWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1, model, data);
// 2: The input-to-forget weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, inputSize] specifying input-to-forget part of weights for fully-connected layer inside the
// LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin inputToForgetWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 2, model, data);
// 3: The input-to-cell weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, inputSize] specifying input-to-cell part of weights for fully-connected layer inside the
// LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin inputToCellWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 3, model, data);
// 4: The input-to-output weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, inputSize] specifying input-to-output part of weights for fully-connected layer inside the
// LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin inputToOutputWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 4, model, data);
// 5: The recurrent-to-input weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, outputSize] specifying recurrent-to-input part of weights for fully-connected layer inside
// the LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin recurrentToInputWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 5, model, data);
// 6: The recurrent-to-forget weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, outputSize] specifying recurrent-to-forget part of weights for fully-connected layer inside
// the LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin recurrentToForgetWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 6, model, data);
// 7: The recurrent-to-cell weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, outputSize] specifying recurrent-to-cell part of weights for fully-connected layer inside
// the LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin recurrentToCellWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 7, model, data);
// 8: The recurrent-to-output weights. A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape
// [outputSize, outputSize] specifying recurrent-to-output part of weights for fully-connected layer inside
// the LSTM cell. Quantization zero point and scale must be the same across all the weights.
const ConstTensorPin recurrentToOutputWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 8, model, data);
// 9: The input gate bias. A 1-D tensor of type ANEURALNETWORKS_TENSOR_INT32 and shape [outputSize] specifying the
// bias for the fully-connected layer inside the LSTM cell. Bias is quantized with scale being a product
// of input and weights scales and zeroPoint equal to 0.
const ConstTensorPin inputGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 9, model, data);
// 10: The forget gate bias. A 1-D tensor of type ANEURALNETWORKS_TENSOR_INT32 and shape [outputSize] specifying
// the bias for the fully-connected layer inside the LSTM cell. Bias is quantized with scale being a product
// of input and weights scales and zeroPoint equal to 0.
const ConstTensorPin forgetGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 10, model, data);
// 11:The cell bias. A 1-D tensor of type ANEURALNETWORKS_TENSOR_INT32 and shape [outputSize] specifying the bias
// for the fully-connected layer inside the LSTM cell. Bias is quantized with scale being a product of input
// and weights scales and zeroPoint equal to 0.
const ConstTensorPin cellBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 11, model, data);
// 12:The output gate bias. A 1-D tensor of type ANEURALNETWORKS_TENSOR_INT32 and shape [outputSize] specifying
// the bias for the fully-connected layer inside the LSTM cell. Bias is quantized with scale being a product
// of input and weights scales and zeroPoint equal to 0.
const ConstTensorPin outputGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 12, model, data);
if (!inputToInputWeightsPin.IsValid() ||
!inputToForgetWeightsPin.IsValid() ||
!inputToCellWeightsPin.IsValid() ||
!inputToOutputWeightsPin.IsValid() ||
!recurrentToInputWeightsPin.IsValid() ||
!recurrentToForgetWeightsPin.IsValid() ||
!recurrentToCellWeightsPin.IsValid() ||
!recurrentToOutputWeightsPin.IsValid() ||
!inputGateBiasPin.IsValid() ||
!forgetGateBiasPin.IsValid() ||
!cellBiasPin.IsValid() ||
!outputGateBiasPin.IsValid())
{
return Fail("%s: Operation has invalid tensor inputs", __func__);
}
// Outputs:
// 0: The cell state: A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT16_SYMM and shape [numBatches, outputSize]
// which contains a cell state from the current time step. Tensor is quantized using a quantization range
// of -2^4, 2^4 * 32767/32768.
const HalOperand* cellStateOut = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!cellStateOut)
{
return Fail("%s: Could not read output 0: cellStateOut", __func__);
}
// 1: The output: A 2-D tensor of type ANEURALNETWORKS_TENSOR_QUANT8_ASYMM and shape [numBathes, outputSize] which
// contains the output value. Tensor is quantized with a fixed quantization range of -1, 127/128.
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 1, model);
if (!output)
{
return Fail("%s: Could not read output 1: output", __func__);
}
// Inputs
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& previousCellStateInInfo = previousCellStateIn.GetTensorInfo();
const TensorInfo& previousOutputInInfo = previousOutputIn.GetTensorInfo();
// Outputs
const TensorInfo& cellStateOutInfo = GetTensorInfoForOperand(*cellStateOut);
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// Dynamic tensors currently not supported
if (IsDynamicTensor(cellStateOutInfo) || IsDynamicTensor(outputInfo))
{
return Fail("%s: Dynamic output tensors are not supported", __func__);
}
QuantizedLstmInputParams params;
params.m_InputToInputWeights = inputToInputWeightsPin.GetConstTensorPtr();
params.m_InputToForgetWeights = inputToForgetWeightsPin.GetConstTensorPtr();
params.m_InputToCellWeights = inputToCellWeightsPin.GetConstTensorPtr();
params.m_InputToOutputWeights = inputToOutputWeightsPin.GetConstTensorPtr();
params.m_RecurrentToInputWeights = recurrentToInputWeightsPin.GetConstTensorPtr();
params.m_RecurrentToForgetWeights = recurrentToForgetWeightsPin.GetConstTensorPtr();
params.m_RecurrentToCellWeights = recurrentToCellWeightsPin.GetConstTensorPtr();
params.m_RecurrentToOutputWeights = recurrentToOutputWeightsPin.GetConstTensorPtr();
params.m_InputGateBias = inputGateBiasPin.GetConstTensorPtr();
params.m_ForgetGateBias = forgetGateBiasPin.GetConstTensorPtr();
params.m_CellBias = cellBiasPin.GetConstTensorPtr();
params.m_OutputGateBias = outputGateBiasPin.GetConstTensorPtr();
QuantizedLstmInputParamsInfo paramsInfo;
paramsInfo.m_InputToInputWeights = &(params.m_InputToInputWeights->GetInfo());
paramsInfo.m_InputToForgetWeights = &(params.m_InputToForgetWeights->GetInfo());
paramsInfo.m_InputToCellWeights = &(params.m_InputToCellWeights->GetInfo());
paramsInfo.m_InputToOutputWeights = &(params.m_InputToOutputWeights->GetInfo());
paramsInfo.m_RecurrentToInputWeights = &(params.m_RecurrentToInputWeights->GetInfo());
paramsInfo.m_RecurrentToForgetWeights = &(params.m_RecurrentToForgetWeights->GetInfo());
paramsInfo.m_RecurrentToCellWeights = &(params.m_RecurrentToCellWeights->GetInfo());
paramsInfo.m_RecurrentToOutputWeights = &(params.m_RecurrentToOutputWeights->GetInfo());
paramsInfo.m_InputGateBias = &(params.m_InputGateBias->GetInfo());
paramsInfo.m_ForgetGateBias = &(params.m_ForgetGateBias->GetInfo());
paramsInfo.m_CellBias = &(params.m_CellBias->GetInfo());
paramsInfo.m_OutputGateBias = &(params.m_OutputGateBias->GetInfo());
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsQuantizedLstmSupported,
data.m_Backends,
isSupported,
inputInfo,
previousCellStateInInfo,
previousOutputInInfo,
cellStateOutInfo,
outputInfo,
paramsInfo);
};
bool isDynamic = false;
if (!IsDynamicTensor(cellStateOutInfo) &&
!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isDynamic = true;
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddQuantizedLstmLayer(params, "QuantizedLstm");
input.Connect(layer->GetInputSlot(0));
previousCellStateIn.Connect(layer->GetInputSlot(1));
previousOutputIn.Connect(layer->GetInputSlot(2));
if (!isDynamic)
{
return (SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, 0, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 1, *layer, 1, model, data));
}
else
{
return (SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, 0, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(
operation, 1, *layer, 1, model, data, nullptr, validateFunc, ActivationFn::kActivationNone, true));
}
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertResize(const HalOperation& operation,
const HalModel& model,
ConversionData& data,
ResizeMethod resizeMethod)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertResize()");
ALOGV("resizeMethod = %s", GetResizeMethodAsCString(resizeMethod));
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Could not read input 0", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
ResizeDescriptor descriptor;
descriptor.m_Method = resizeMethod;
descriptor.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 3, model, data);
HalOperandType operandType1;
HalOperandType operandType2;
if (!GetOperandType<HalPolicy>(operation, 1, model, operandType1) ||
!GetOperandType<HalPolicy>(operation, 2, model, operandType2))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
if (operandType1 != operandType2)
{
return Fail("%s: Operation has invalid inputs. Type of input 1 and 2 should be the same", __func__);
}
if (operandType1 == HalOperandType::INT32)
{
// Case 1: resizing by shape
int32_t targetWidth = 0;
int32_t targetHeight = 0;
if (!GetInputInt32<HalPolicy>(operation, 1, targetWidth, model, data) ||
!GetInputInt32<HalPolicy>(operation, 2, targetHeight, model, data))
{
return Fail("%s: Operation has invalid inputs for resizing by shape", __func__);
}
if (targetWidth < 0 || targetHeight < 0)
{
return Fail("%s: Operation has invalid inputs for resizing by shape. "
"Target width/height cannot be < 0", __func__);
}
descriptor.m_TargetWidth = static_cast<uint32_t>(targetWidth);
descriptor.m_TargetHeight = static_cast<uint32_t>(targetHeight);
}
else if (operandType1 == HalOperandType::FLOAT32)
{
// Case 2: resizing by scale
float widthScale = 1.0f;
float heightScale = 1.0f;
if (!GetInputFloat32<HalPolicy>(operation, 1, widthScale, model, data) ||
!GetInputFloat32<HalPolicy>(operation, 2, heightScale, model, data))
{
return Fail("%s: Operation has invalid inputs for resizing by scale", __func__);
}
const TensorShape& inputShape = inputInfo.GetShape();
armnnUtils::DataLayoutIndexed dataLayoutIndexed(descriptor.m_DataLayout);
float width = inputShape[dataLayoutIndexed.GetWidthIndex()];
float height = inputShape[dataLayoutIndexed.GetHeightIndex()];
descriptor.m_TargetWidth = std::floor(width * widthScale);
descriptor.m_TargetHeight = std::floor(height * heightScale);
}
else if (operandType1 == HalOperandType::FLOAT16)
{
Half widthScale;
Half heightScale;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT16, widthScale, model, data) ||
!GetInputScalar<HalPolicy>(operation, 2, HalOperandType::FLOAT16, heightScale, model, data))
{
return Fail("%s: Operation has invalid inputs for resizing by scale", __func__);
}
const TensorShape& inputShape = inputInfo.GetShape();
armnnUtils::DataLayoutIndexed dataLayoutIndexed(descriptor.m_DataLayout);
Half width = static_cast<Half>(inputShape[dataLayoutIndexed.GetWidthIndex()]);
Half height = static_cast<Half>(inputShape[dataLayoutIndexed.GetHeightIndex()]);
descriptor.m_TargetWidth = std::floor(width * widthScale);
descriptor.m_TargetHeight = std::floor(height * heightScale);
}
else
{
return Fail("%s: Operand has invalid data type for resizing by scale", __func__);
}
descriptor.m_AlignCorners = GetOptionalBool<HalPolicy>(operation, 4, model, data);
descriptor.m_HalfPixelCenters = GetOptionalBool<HalPolicy>(operation, 5, model, data);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsResizeSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
descriptor);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddResizeLayer(descriptor);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertSpaceToDepth(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertSpaceToDepth()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid() )
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
unsigned int rank = inputInfo.GetNumDimensions();
if (rank != 4)
{
return Fail("%s: Only inputs with rank 4 are supported", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
SpaceToDepthDescriptor desc;
GetInputScalar<HalPolicy>(operation, 1, HalOperandType::INT32, desc.m_BlockSize, model, data);
if (desc.m_BlockSize <= 1)
{
return Fail("%s: Block size must be at least 1 in all dimensions");
}
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 2, model, data);
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsSpaceToDepthSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
desc);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* const layer = data.m_Network->AddSpaceToDepthLayer(desc);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertSoftmax(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertSoftmax()");
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* outputOperand = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!outputOperand)
{
return Fail("%s: Operation has no outputs", __func__);
}
const TensorInfo& outputInfo = GetTensorInfoForOperand(*outputOperand);
SoftmaxDescriptor desc;
HalOperandType outputType = outputOperand->type;
// Read beta value
if (outputType == HalOperandType::TENSOR_FLOAT16)
{
Half value;
if (!GetInputScalar<HalPolicy>(operation, 1, HalOperandType::FLOAT16, value, model, data))
{
return Fail("%s: Operation has invalid inputs %d", __func__, outputType);
}
desc.m_Beta = static_cast<float>(value);
}
else
{
if (!GetInputFloat32<HalPolicy>(operation, 1, desc.m_Beta, model, data))
{
return Fail("%s: Operation has invalid inputs %d", __func__, outputType);
}
}
if (operation.inputs.size() > 2 && !GetInputScalar<HalPolicy>(operation,
2,
HalOperandType::INT32,
desc.m_Axis,
model,
data))
{
return Fail("%s: Operation has invalid inputs", __func__);
}
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsSoftmaxSupported,
data.m_Backends,
isSupported,
input.GetTensorInfo(),
outputInfo,
desc);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* layer = data.m_Network->AddSoftmaxLayer(desc);
assert(layer != nullptr);
input.Connect(layer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, model, data, nullptr, validateFunc);
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertLstm(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
ALOGV("HalPolicy::ConvertLstm()");
// Inputs:
// 00: The input: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, input_size], where
// “batch_size” corresponds to the batching dimension, and “input_size” is the size of the input.
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Could not read input 0: input", __func__);
}
// 18: The output state: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, output_size].
LayerInputHandle outputStateIn = ConvertToLayerInputHandle<HalPolicy>(operation, 18, model, data);
if (!outputStateIn.IsValid())
{
return Fail("%s: Could not read input 18: outputStateIn", __func__);
}
// 19: The cell state: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, num_units].
LayerInputHandle cellStateIn = ConvertToLayerInputHandle<HalPolicy>(operation, 19, model, data);
if (!cellStateIn.IsValid())
{
return Fail("%s: Could not read input 19: cellStateIn", __func__);
}
// Get the mandatory input tensors:
// 02: The input-to-forget weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, input_size].
const ConstTensorPin inputToForgetWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 2));
// 03: The input-to-cell weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, input_size].
const ConstTensorPin inputToCellWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 3));
// 04: The input-to-output weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, input_size].
const ConstTensorPin inputToOutputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 4));
// 06: The recurrent-to-forget weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, output_size].
const ConstTensorPin recurrentToForgetWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 6));
// 07: The recurrent-to-cell weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, output_size].
const ConstTensorPin recurrentToCellWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 7));
// 08: The recurrent-to-output weights: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, output_size].
const ConstTensorPin recurrentToOutputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 8));
// 13: The forget gate bias: A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin forgetGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 13, model, data);
// 14: The cell bias: A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin cellBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 14, model, data);
// 15: The output gate bias: A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin outputGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 15, model, data);
if (!inputToForgetWeightsPin.IsValid() ||
!inputToCellWeightsPin.IsValid() ||
!inputToOutputWeightsPin.IsValid() ||
!recurrentToForgetWeightsPin.IsValid() ||
!recurrentToCellWeightsPin.IsValid() ||
!recurrentToOutputWeightsPin.IsValid() ||
!forgetGateBiasPin.IsValid() ||
!cellBiasPin.IsValid() ||
!outputGateBiasPin.IsValid())
{
return Fail("%s: Operation has invalid tensor inputs", __func__);
}
// Get the optional input tensors:
// 01: The input-to-input weights: Optional. A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, input_size], where “num_units” corresponds to the number of cell units.
const ConstTensorPin inputToInputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 1, true));
// 05: The recurrent-to-input weights: Optional. A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [num_units, output_size], where “output_size” corresponds to either the number of cell units (i.e.,
// “num_units”), or the second dimension of the “projection_weights”, if defined.
const ConstTensorPin recurrentToInputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 5, true));
// 09: The cell-to-input weights: Optional. A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin cellToInputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 9, true));
// 10: The cell-to-forget weights: Optional. A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin cellToForgetWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 10, true));
// 11: The cell-to-output weights: Optional. A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin cellToOutputWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 11, true));
// 12: The input gate bias: Optional. A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [num_units].
const ConstTensorPin inputGateBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
12,
model,
data,
g_DontPermute,
nullptr,
true);
// 16: The projection weights: Optional. A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape
// [output_size, num_units].
const ConstTensorPin projectionWeightsPin =
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 16, true));
// 17: The projection bias: Optional. A 1-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [output_size].
const ConstTensorPin projectionBiasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
17,
model,
data,
g_DontPermute,
nullptr,
true);
if ((!inputToInputWeightsPin.IsValid() && !inputToInputWeightsPin.IsOptional()) ||
(!recurrentToInputWeightsPin.IsValid() && !recurrentToInputWeightsPin.IsOptional()) ||
(!cellToInputWeightsPin.IsValid() && !cellToInputWeightsPin.IsOptional()) ||
(!cellToForgetWeightsPin.IsValid() && !cellToForgetWeightsPin.IsOptional()) ||
(!cellToOutputWeightsPin.IsValid() && !cellToOutputWeightsPin.IsOptional()) ||
(!inputGateBiasPin.IsValid() && !inputGateBiasPin.IsOptional()) ||
(!projectionWeightsPin.IsValid() && !projectionWeightsPin.IsOptional()) ||
(!projectionBiasPin.IsValid() && !projectionBiasPin.IsOptional()))
{
return Fail("%s: Operation has invalid tensor inputs", __func__);
}
// Get the mandatory input scalars (actually 1-D tensors of size 1):
// 20: The activation function: A value indicating the activation function:
// 0: None; 1: Relu; 3: Relu6; 4: Tanh; 6: Sigmoid.
// 21: The clipping threshold: for the cell state, such that values are bound within [-cell_clip, cell_clip].
// If set to 0.0 then clipping is disabled.
// 22: The clipping threshold: for the output from the projection layer, such that values are bound within
// [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled.
ActivationFn activation;
float cellClip;
float projClip;
if (!GetInputActivationFunctionFromTensor<HalPolicy>(operation, 20, activation, model, data) ||
!GetInputScalar<HalPolicy>(operation, 21, HalOperandType::FLOAT32, cellClip, model, data) ||
!GetInputScalar<HalPolicy>(operation, 22, HalOperandType::FLOAT32, projClip, model, data))
{
return Fail("%s: Operation has invalid scalar inputs", __func__);
}
// Get the normalization tensors
// 23: The input layer normalization weights. A 1-D tensor of shape [num_units].
// Used to rescale normalized inputs to activation at input gate.
const ConstTensorPin inputLayerNormWeightsPin
(DequantizeAndMakeConstTensorPin<HalPolicy>(operation, model, data, 23, true));
// 24: The forget layer normalization weights. A 1-D tensor of shape [num_units].
// Used to rescale normalized inputs to activation at forget gate.
const ConstTensorPin forgetLayerNormWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
24,
model,
data,
g_DontPermute,
nullptr,
true);
// 25: The cell layer normalization weights. A 1-D tensor of shape [num_units].
// Used to rescale normalized inputs to activation at cell gate.
const ConstTensorPin cellLayerNormWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
25,
model,
data,
g_DontPermute,
nullptr,
true);
// 26: The output layer normalization weights. A 1-D tensor of shape [num_units].
// Used to rescale normalized inputs to activation at output gate.
const ConstTensorPin outputLayerNormWeightsPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation,
26,
model,
data,
g_DontPermute,
nullptr,
true);
// Outputs:
// 00: The scratch buffer: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, num_units * 4]
// with CIFG, or [batch_size, num_units * 3] without CIFG.
const HalOperand* scratchBuffer = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!scratchBuffer)
{
return Fail("%s: Could not read output 0: scratchBuffer", __func__);
}
// 01: The output state (out): A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, output_size].
const HalOperand* outputStateOut = GetOutputOperand<HalPolicy>(operation, 1, model);
if (!outputStateOut)
{
return Fail("%s: Could not read output 1: outputStateOut", __func__);
}
// 02: The cell state (out): A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, num_units].
const HalOperand* cellStateOut = GetOutputOperand<HalPolicy>(operation, 2, model);
if (!cellStateOut)
{
return Fail("%s: Could not read output 2: cellStateOut", __func__);
}
// 03: The output: A 2-D tensor of ANEURALNETWORKS_TENSOR_FLOAT32, of shape [batch_size, output_size]. This is
// effectively the same as the current “output state (out)” value.
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 3, model);
if (!output)
{
return Fail("%s: Could not read output 3: output", __func__);
}
// set the params structure for the AddLstmLayer call
LstmInputParams params;
params.m_InputToInputWeights = inputToInputWeightsPin.GetConstTensorPtr();
params.m_InputToForgetWeights = inputToForgetWeightsPin.GetConstTensorPtr();
params.m_InputToCellWeights = inputToCellWeightsPin.GetConstTensorPtr();
params.m_InputToOutputWeights = inputToOutputWeightsPin.GetConstTensorPtr();
params.m_RecurrentToInputWeights = recurrentToInputWeightsPin.GetConstTensorPtr();
params.m_RecurrentToForgetWeights = recurrentToForgetWeightsPin.GetConstTensorPtr();
params.m_RecurrentToCellWeights = recurrentToCellWeightsPin.GetConstTensorPtr();
params.m_RecurrentToOutputWeights = recurrentToOutputWeightsPin.GetConstTensorPtr();
params.m_CellToInputWeights = cellToInputWeightsPin.GetConstTensorPtr();
params.m_CellToForgetWeights = cellToForgetWeightsPin.GetConstTensorPtr();
params.m_CellToOutputWeights = cellToOutputWeightsPin.GetConstTensorPtr();
params.m_InputGateBias = inputGateBiasPin.GetConstTensorPtr();
params.m_ForgetGateBias = forgetGateBiasPin.GetConstTensorPtr();
params.m_CellBias = cellBiasPin.GetConstTensorPtr();
params.m_OutputGateBias = outputGateBiasPin.GetConstTensorPtr();
params.m_ProjectionWeights = projectionWeightsPin.GetConstTensorPtr();
params.m_ProjectionBias = projectionBiasPin.GetConstTensorPtr();
params.m_InputLayerNormWeights = inputLayerNormWeightsPin.GetConstTensorPtr();
params.m_ForgetLayerNormWeights = forgetLayerNormWeightsPin.GetConstTensorPtr();
params.m_CellLayerNormWeights = cellLayerNormWeightsPin.GetConstTensorPtr();
params.m_OutputLayerNormWeights = outputLayerNormWeightsPin.GetConstTensorPtr();
// set the layer descriptor
LstmDescriptor desc;
desc.m_ActivationFunc = activation;
desc.m_ClippingThresCell = cellClip;
desc.m_ClippingThresProj = projClip;
desc.m_CifgEnabled = (params.m_InputToInputWeights == nullptr ||
params.m_RecurrentToInputWeights == nullptr ||
params.m_InputGateBias == nullptr);
desc.m_PeepholeEnabled = (params.m_CellToForgetWeights != nullptr ||
params.m_CellToOutputWeights != nullptr);
desc.m_ProjectionEnabled = (params.m_ProjectionWeights != nullptr);
desc.m_LayerNormEnabled = (params.m_InputLayerNormWeights != nullptr ||
params.m_ForgetLayerNormWeights != nullptr ||
params.m_CellLayerNormWeights != nullptr ||
params.m_OutputLayerNormWeights != nullptr);
// validate the optional input groups
if (desc.m_CifgEnabled &&
(params.m_InputToInputWeights != nullptr ||
params.m_RecurrentToInputWeights != nullptr ||
params.m_InputGateBias != nullptr))
{
return Fail("%s: All, or none, of input-to-input weights, recurrent-to-input weights,"
" and input gate bias must be provided", __func__);
}
if (!desc.m_ProjectionEnabled && params.m_ProjectionBias != nullptr)
{
return Fail("%s: projection bias should not be provided without projection weights", __func__);
}
if (desc.m_PeepholeEnabled &&
(params.m_CellToForgetWeights == nullptr ||
params.m_CellToOutputWeights == nullptr ||
(!desc.m_CifgEnabled && params.m_CellToInputWeights == nullptr)))
{
return Fail("%s: All, or none, of cell-to-forget weights and cell-to-output weights must be provided"
" and, if CIFG is not enabled, cell-to-input weights must also be provided", __func__);
}
if (desc.m_LayerNormEnabled &&
(params.m_ForgetLayerNormWeights == nullptr ||
params.m_CellLayerNormWeights == nullptr ||
params.m_OutputLayerNormWeights == nullptr ||
(!desc.m_CifgEnabled && params.m_InputLayerNormWeights == nullptr)))
{
return Fail("%s: All, or none, of forget-norm weights, cell-norm weights and output-norm weights must be"
" provided and, if CIFG is not enabled, input-norm weights must also be provided", __func__);
}
// Check if the layer is supported
// Inputs
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputStateInInfo = outputStateIn.GetTensorInfo();
const TensorInfo& cellStateInInfo = cellStateIn.GetTensorInfo();
// Outputs
const TensorInfo& scratchBufferInfo = GetTensorInfoForOperand(*scratchBuffer);
const TensorInfo& outputStateOutInfo = GetTensorInfoForOperand(*outputStateOut);
const TensorInfo& cellStateOutInfo = GetTensorInfoForOperand(*cellStateOut);
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// Basic parameters
LstmInputParamsInfo paramsInfo;
paramsInfo.m_InputToForgetWeights = &(params.m_InputToForgetWeights->GetInfo());
paramsInfo.m_InputToCellWeights = &(params.m_InputToCellWeights->GetInfo());
paramsInfo.m_InputToOutputWeights = &(params.m_InputToOutputWeights->GetInfo());
paramsInfo.m_RecurrentToForgetWeights = &(params.m_RecurrentToForgetWeights->GetInfo());
paramsInfo.m_RecurrentToCellWeights = &(params.m_RecurrentToCellWeights->GetInfo());
paramsInfo.m_RecurrentToOutputWeights = &(params.m_RecurrentToOutputWeights->GetInfo());
paramsInfo.m_ForgetGateBias = &(params.m_ForgetGateBias->GetInfo());
paramsInfo.m_CellBias = &(params.m_CellBias->GetInfo());
paramsInfo.m_OutputGateBias = &(params.m_OutputGateBias->GetInfo());
// Optional parameters
if (!desc.m_CifgEnabled)
{
paramsInfo.m_InputToInputWeights = &(params.m_InputToInputWeights->GetInfo());
paramsInfo.m_RecurrentToInputWeights = &(params.m_RecurrentToInputWeights->GetInfo());
if (params.m_CellToInputWeights != nullptr)
{
paramsInfo.m_CellToInputWeights = &(params.m_CellToInputWeights->GetInfo());
}
paramsInfo.m_InputGateBias = &(params.m_InputGateBias->GetInfo());
}
if (desc.m_ProjectionEnabled)
{
paramsInfo.m_ProjectionWeights = &(params.m_ProjectionWeights->GetInfo());
if (params.m_ProjectionBias != nullptr)
{
paramsInfo.m_ProjectionBias = &(params.m_ProjectionBias->GetInfo());
}
}
if (desc.m_PeepholeEnabled)
{
paramsInfo.m_CellToForgetWeights = &(params.m_CellToForgetWeights->GetInfo());
paramsInfo.m_CellToOutputWeights = &(params.m_CellToOutputWeights->GetInfo());
}
if (desc.m_LayerNormEnabled)
{
if(!desc.m_CifgEnabled)
{
paramsInfo.m_InputLayerNormWeights = &(params.m_InputLayerNormWeights->GetInfo());
}
paramsInfo.m_ForgetLayerNormWeights = &(params.m_ForgetLayerNormWeights->GetInfo());
paramsInfo.m_CellLayerNormWeights = &(params.m_CellLayerNormWeights->GetInfo());
paramsInfo.m_OutputLayerNormWeights = &(params.m_OutputLayerNormWeights->GetInfo());
}
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsLstmSupported,
data.m_Backends,
isSupported,
inputInfo,
outputStateInInfo,
cellStateInInfo,
scratchBufferInfo,
outputStateOutInfo,
cellStateOutInfo,
outputInfo,
desc,
paramsInfo);
};
bool isDynamic = false;
if (!IsDynamicTensor(outputStateOutInfo) &&
!IsDynamicTensor(scratchBufferInfo) &&
!IsDynamicTensor(cellStateOutInfo) &&
!IsDynamicTensor(outputInfo))
{
validateFunc(outputInfo, isSupported);
}
else
{
isDynamic = true;
isSupported = AreDynamicTensorsSupported();
}
if (!isSupported)
{
return false;
}
// Add the layer
IConnectableLayer* layer = data.m_Network->AddLstmLayer(desc, params, "Lstm");
input.Connect(layer->GetInputSlot(0));
outputStateIn.Connect(layer->GetInputSlot(1));
cellStateIn.Connect(layer->GetInputSlot(2));
if (!isDynamic)
{
return (
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, 0, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 1, *layer, 1, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 2, *layer, 2, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 3, *layer, 3, model, data));
}
else
{
return (
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *layer, 0, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 1, *layer, 1, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 2, *layer, 2, model, data) &&
SetupAndTrackLayerOutputSlot<HalPolicy>(
operation, 3, *layer, 3, model, data, nullptr, validateFunc, ActivationFn::kActivationNone, true));
}
}
template<typename HalPolicy,
typename HalOperation = typename HalPolicy::Operation,
typename HalModel = typename HalPolicy::Model>
bool ConvertTransposeConv2d(const HalOperation& operation, const HalModel& model, ConversionData& data)
{
using HalOperand = typename HalPolicy::Operand;
using HalOperandType = typename HalPolicy::OperandType;
LayerInputHandle input = ConvertToLayerInputHandle<HalPolicy>(operation, 0, model, data);
if (!input.IsValid())
{
return Fail("%s: Operation has invalid inputs", __func__);
}
const HalOperand* output = GetOutputOperand<HalPolicy>(operation, 0, model);
if (!output)
{
return Fail("%s: Could not read output 0", __func__);
}
const TensorInfo& inputInfo = input.GetTensorInfo();
const TensorInfo& outputInfo = GetTensorInfoForOperand(*output);
// ArmNN does not currently support non-fixed weights or bias
// Find the shape of the weights tensor. In AndroidNN this will be [ 1, H, W, I * M ]
const HalOperand* weightsOperand = GetInputOperand<HalPolicy>(operation, 1, model);
if (weightsOperand == nullptr)
{
return Fail("%s: Operand is invalid", __func__);
}
TransposeConvolution2dDescriptor desc;
desc.m_DataLayout = DataLayout::NHWC;
// Determine whether padding is implicit or explicit
bool implicitPadding = operation.inputs.size() == 9;
if (implicitPadding )
{
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 8, model, data);
}
else
{
desc.m_DataLayout = OptionalDataLayout<HalPolicy>(operation, 10, model, data);
}
armnnUtils::DataLayoutIndexed dataLayoutIndexed(desc.m_DataLayout);
unsigned int widthIndex = dataLayoutIndexed.GetWidthIndex();
unsigned int heightIndex = dataLayoutIndexed.GetHeightIndex();
const PermutationVector OHWIToOIHW = {0, 2, 3, 1};
// The shape of the weight is [depth_out, filter_height, filter_width, depth_in].
// We have to permute it to OIHW if the data layout is NCHW.
const ConstTensorPin weightsPin = (desc.m_DataLayout == DataLayout::NCHW) ?
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1,
model, data, OHWIToOIHW) :
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 1, model, data);
// Bias is a 1D tensor
const ConstTensorPin biasPin =
ConvertOperationInputToConstTensorPin<HalPolicy>(operation, 2, model, data);
if (!weightsPin.IsValid())
{
return Fail("%s: Operation has invalid weights", __func__);
}
if (!biasPin.IsValid())
{
return Fail("%s: Operation has invalid biases", __func__);
}
ConstTensor weights = weightsPin.GetConstTensor();
ConstTensor bias = biasPin.GetConstTensor();
SanitizeBiasQuantizationScale(bias.GetInfo(), weights.GetInfo(), inputInfo);
ActivationFn activation;
if (implicitPadding)
{
int32_t strideX{0};
int32_t strideY{0};
int32_t padLeft{0};
int32_t padRight{0};
int32_t padTop{0};
int32_t padBottom{0};
android::nn::PaddingScheme paddingScheme;
if (!GetInputPaddingScheme<HalPolicy>(operation, 4, paddingScheme, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, strideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, strideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 7, activation, model, data))
{
return Fail("%s: Operation has invalid inputs (implicit padding)", __func__);
}
const uint32_t kernelX = weights.GetShape()[widthIndex];
const uint32_t kernelY = weights.GetShape()[heightIndex];
// If output shape has been specified as a parameter then extract it and make it available.
const HalOperand* outputShapeOperand = GetInputOperand<HalPolicy>(operation, 3, model, false);
std::vector<int32_t> outputShape;
if ((outputShapeOperand) && (GetTensorInt32Values<HalPolicy>(*outputShapeOperand, outputShape, model, data)))
{
// Change from signed to unsigned int to store in TransposeConvolution2dDescriptor.
for (int dimension : outputShape)
{
desc.m_OutputShape.push_back(static_cast<unsigned int>(dimension));
}
desc.m_OutputShapeEnabled = true;
}
uint32_t outputX;
uint32_t outputY;
if (IsDynamicTensor(outputInfo))
{
if (outputShape.size() == 0)
{
return Fail("%s: Padding sizes cannot be inferred", __func__);
}
outputX = outputShape[widthIndex];
outputY = outputShape[heightIndex];
}
else
{
outputX = outputInfo.GetShape()[widthIndex];
outputY = outputInfo.GetShape()[heightIndex];
}
CalcPaddingTransposeConv(outputX, kernelX, strideX, padLeft, padRight, paddingScheme);
CalcPaddingTransposeConv(outputY, kernelY, strideY, padTop, padBottom, paddingScheme);
// NOTE: The Android NN API allows for negative padding values in TransposeConv2d,
// but Arm NN only supports values >= 0
if (padLeft < 0 || padRight < 0 || padTop < 0 || padBottom < 0)
{
return Fail("%s: Negative padding values are not supported", __func__);
}
desc.m_StrideX = armnn::numeric_cast<uint32_t>(strideX);
desc.m_StrideY = armnn::numeric_cast<uint32_t>(strideY);
desc.m_PadLeft = armnn::numeric_cast<uint32_t>(padLeft);
desc.m_PadRight = armnn::numeric_cast<uint32_t>(padRight);
desc.m_PadTop = armnn::numeric_cast<uint32_t>(padTop);
desc.m_PadBottom = armnn::numeric_cast<uint32_t>(padBottom);
}
else if (operation.inputs.size() == 11)
{
// explicit padding
if (!GetInputScalar<HalPolicy>(operation, 3, HalOperandType::INT32, desc.m_PadLeft, model, data) ||
!GetInputScalar<HalPolicy>(operation, 4, HalOperandType::INT32, desc.m_PadRight, model, data) ||
!GetInputScalar<HalPolicy>(operation, 5, HalOperandType::INT32, desc.m_PadTop, model, data) ||
!GetInputScalar<HalPolicy>(operation, 6, HalOperandType::INT32, desc.m_PadBottom, model, data) ||
!GetInputScalar<HalPolicy>(operation, 7, HalOperandType::INT32, desc.m_StrideX, model, data) ||
!GetInputScalar<HalPolicy>(operation, 8, HalOperandType::INT32, desc.m_StrideY, model, data) ||
!GetInputActivationFunction<HalPolicy>(operation, 9, activation, model, data))
{
return Fail("%s: Operation has invalid inputs (explicit padding)", __func__);
}
}
else
{
return Fail("%s: Unsupported number of operation inputs", __func__);
}
desc.m_BiasEnabled = true;
Optional<TensorInfo> biases(bias.GetInfo());
bool isSupported = false;
auto validateFunc = [&](const armnn::TensorInfo& outputInfo, bool& isSupported)
{
FORWARD_LAYER_SUPPORT_FUNC(__func__,
IsTransposeConvolution2dSupported,
data.m_Backends,
isSupported,
inputInfo,
outputInfo,
desc,
weights.GetInfo(),
biases);
};
if(IsDynamicTensor(outputInfo))
{
isSupported = AreDynamicTensorsSupported();
}
else
{
validateFunc(outputInfo, isSupported);
}
if (!isSupported)
{
return false;
}
IConnectableLayer* startLayer =
data.m_Network->AddTransposeConvolution2dLayer(desc, weights, Optional<ConstTensor>(bias));
if (!startLayer)
{
return Fail("%s: AddTransposeConvolution2dLayer failed", __func__);
}
input.Connect(startLayer->GetInputSlot(0));
return SetupAndTrackLayerOutputSlot<HalPolicy>(operation, 0, *startLayer, model,
data, nullptr, validateFunc, activation);
}
} // armnn_driver namespace