| // |
| // Copyright © 2023 Arm Ltd and Contributors. All rights reserved. |
| // SPDX-License-Identifier: MIT |
| // |
| |
| #pragma once |
| |
| namespace |
| { |
| |
| armnn::INetworkPtr CreateReverseV2Network(const armnn::TensorInfo& inputInfo, |
| const armnn::TensorInfo& axisInfo, |
| const armnn::TensorInfo& outputInfo, |
| const std::vector<int32_t>& axisData) |
| { |
| using namespace armnn; |
| |
| INetworkPtr network(INetwork::Create()); |
| |
| IConnectableLayer* inputLayer = network->AddInputLayer(0, "input"); |
| IConnectableLayer* axisLayer = network->AddConstantLayer(ConstTensor(axisInfo, axisData)); |
| IConnectableLayer* reverseLayer = network->AddReverseV2Layer("reverseV2"); |
| IConnectableLayer* outputLayer = network->AddOutputLayer(0, "output"); |
| |
| Connect(inputLayer, reverseLayer, inputInfo, 0, 0); |
| Connect(axisLayer, reverseLayer, axisInfo, 0, 1); |
| Connect(reverseLayer, outputLayer, outputInfo, 0, 0); |
| |
| return network; |
| } |
| |
| template <armnn::DataType ArmnnType, typename T = armnn::ResolveType<ArmnnType>> |
| void ReverseV2EndToEnd(const std::vector<BackendId>& backends) |
| { |
| float qScale = 1.0f; |
| int32_t qOffset = 0; |
| bool qConst = true; |
| |
| TensorInfo inputInfo ( { 2, 3, 4 }, ArmnnType, qScale, qOffset, qConst); |
| TensorInfo axisInfo ( { 1 } , DataType::Signed32, qScale, qOffset, qConst); |
| TensorInfo outputInfo ( { 2, 3, 4 }, ArmnnType, qScale, qOffset); |
| |
| std::vector<T> inputData = armnnUtils::QuantizedVector<T>({ |
| 1, 2, 3, 4, |
| 5, 6 , 7, 8, |
| 9, 10, 11, 12, |
| 13, 14, 15, 16, |
| 17, 18, 19, 20, |
| 21, 22, 23, 24 |
| }, qScale, qOffset); |
| |
| std::vector<int32_t> axisData = { 1 }; |
| |
| std::vector<T> expectedOutputData = armnnUtils::QuantizedVector<T>({ |
| 9, 10, 11, 12, |
| 5, 6, 7 , 8, |
| 1, 2, 3, 4, |
| 21, 22, 23, 24, |
| 17, 18, 19, 20, |
| 13, 14, 15, 16 |
| }, qScale, qOffset); |
| |
| INetworkPtr network = CreateReverseV2Network(inputInfo, axisInfo, outputInfo, axisData); |
| |
| std::map<int, std::vector<T>> inputTensor = { { 0, inputData } }; |
| std::map<int, std::vector<T>> expectedOutputTensor = { { 0, expectedOutputData } }; |
| |
| EndToEndLayerTestImpl<ArmnnType, ArmnnType>(std::move(network), inputTensor, expectedOutputTensor, backends); |
| } |
| |
| } // anonymous namespace |
| |