diff --git a/.jenkins/check/config/filter_pylint.txt b/.jenkins/check/config/filter_pylint.txt
index 0be60f38a67d76e2bcd6217e64e1b5bc6b80c2ba..5c609f01b4fdde746d27d28e9d70f061673e38ae 100644
--- a/.jenkins/check/config/filter_pylint.txt
+++ b/.jenkins/check/config/filter_pylint.txt
@@ -9,3 +9,6 @@
"models/official/cv" "missing-docstring"
"models/official/cv" "c-extension-no-member"
"models/official/nlp/bert_thor/src/bert_model.py" "redefined-outer-name"
+
+# research
+"models/research/cvtmodel" "missing-docstring"
\ No newline at end of file
diff --git a/.jenkins/check/config/whitelizard.txt b/.jenkins/check/config/whitelizard.txt
index cc67c0413ece3e1b81b894af5f3d1f2e5d3b99c2..11170800b5b7c565d37f8b7fa68a9b87ada9fd41 100644
--- a/.jenkins/check/config/whitelizard.txt
+++ b/.jenkins/check/config/whitelizard.txt
@@ -5,3 +5,20 @@
#
models/official/recommend/wide_and_deep/src/wide_and_deep.py:__init__
models/official/recommend/wide_and_deep_multitable/src/wide_and_deep.py:__init__
+models/research/cvtmodel/wide_resnet/src/wide_resnet101_2.py:__init__
+models/research/cvtmodel/resnest/src/resnest200.py:__init__
+models/research/cvtmodel/resnest/src/resnest200.py:construct
+models/research/cvtmodel/resnest/src/resnest101.py:__init__
+models/research/cvtmodel/resnest/src/resnest101.py:construct
+models/research/cvtmodel/resnest/src/resnest50.py:__init__
+models/research/cvtmodel/resnest/src/resnest50.py:construct
+models/research/cvtmodel/resnest/src/resnest269.py:__init__
+models/research/cvtmodel/resnest/src/resnest269.py:construct
+models/research/cvtmodel/densenet/src/densenet169.py:__init__
+models/research/cvtmodel/densenet/src/densenet169.py:construct
+models/research/cvtmodel/densenet/src/densenet201.py:__init__
+models/research/cvtmodel/densenet/src/densenet201.py:construct
+models/research/cvtmodel/densenet/src/densenet121.py:__init__
+models/research/cvtmodel/densenet/src/densenet121.py:construct
+models/research/cvtmodel/densenet/src/densenet161.py:__init__
+models/research/cvtmodel/densenet/src/densenet161.py:construct
diff --git a/research/cvtmodel/README_CN.md b/research/cvtmodel/README_CN.md
new file mode 100644
index 0000000000000000000000000000000000000000..9dc845ee94fc04f9e95a37e13f506104f9b262ae
--- /dev/null
+++ b/research/cvtmodel/README_CN.md
@@ -0,0 +1,36 @@
+# Cvtmodel说明与要求
+
+## Cvtmodel介绍
+
+Cvtmodel文件夹中包含的模型，均是通过Mindconverter直接转换Pytorch所提供的pth模型，并实现了对该网络的Ascend 310推理。
+
+## Cvtmodel使用流程
+
+### 模型下载
+
+以densenet系列网络为例， 首先需要在pytorch提供网页上找到模型的下载方式，按照指南将pth模型文件下载并转化为onnx模型文件。(参考https://pytorch.org/hub/pytorch_vision_densenet/)
+
+### 模型转换
+
+在成功获得onnx文件以后，使用Mindconverter对onnx进行转化，示例如下。
+
+'''shell
+export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=cpp
+mindconverter --model_file ./onnx/densenet-121.onnx --shape 1,3,224,224 --input_nodes images --output_nodes output --output ./
+'''
+
+### 模型推理
+
+Mindconverter转换出ckpt文件后，再使用export.py脚本导出MINDIR文件，根据模型文件夹内readme所写的指令进行Ascend 310推理即可。
+
+## 环境要求
+
+- 硬件（Ascend/GPU）
+- 准备Ascend或GPU处理器搭建硬件环境。
+- 框架
+- [MindSpore](https://www.mindspore.cn/install)
+- [MindInsight] (https://www.mindspore.cn/mindinsight/docs/zh-CN/master/mindinsight_install.html)
+- 如需查看详情，请参见如下资源：
+- [MindSpore教程](https://www.mindspore.cn/tutorials/zh-CN/master/index.html)
+- [MindSpore Python API](https://www.mindspore.cn/docs/api/zh-CN/master/index.html)
+- [MindConverter教程] (https://www.mindspore.cn/mindinsight/docs/zh-CN/master/migrate_3rd_scripts_mindconverter.html)
\ No newline at end of file
diff --git a/research/cvtmodel/densenet/README_CN.md b/research/cvtmodel/densenet/README_CN.md
new file mode 100644
index 0000000000000000000000000000000000000000..9e1e57a7eb1a18b558ea767267942e42acc38d1b
--- /dev/null
+++ b/research/cvtmodel/densenet/README_CN.md
@@ -0,0 +1,126 @@
+# 目录
+
+<!-- TOC -->
+
+- [目录](#目录)
+- [DenseNet描述](#densenet描述)
+- [模型架构](#模型架构)
+- [数据集](#数据集)
+- [环境要求](#环境要求)
+- [脚本说明](#脚本说明)
+ - [脚本及样例代码](#脚本及样例代码)
+ - [导出过程](#导出过程)
+ - [导出](#导出)
+ - [推理过程](#推理过程)
+ - [推理](#推理)
+- [ModelZoo主页](#modelzoo主页)
+
+<!-- /TOC -->
+
+# DenseNet描述
+
+DenseNet是一个基于卷积的神经网络，用于图像分类。有关该模型的描述，可查阅[此论文](https://arxiv.org/abs/1608.06993)。
+本仓库中是基于提供的模型文件，使用MindConverter工具转化出Mindspore框架内的ckpt文件，进行全量推理以验证模型文件精度。
+模型可从网址中下载https://pytorch.org/hub/pytorch_vision_densenet/。
+
+# 模型架构
+
+DenseNet模型支持四种模式：DenseNet-121, DenseNet-161，DenseNet-169, DenseNet-201。数字表示网络中包含的卷积层数量。
+
+# 数据集
+
+DenseNet使用的数据集： ImageNet
+
+数据集的默认配置如下：
+
+- 测试数据集预处理：
+ - 图像的输入尺寸：224\*224（将图像缩放到256\*256，然后在中央区域裁剪图像）
+ - 根据平均值和标准偏差对输入图像进行归一化
+
+# 环境要求
+
+- 硬件（Ascend/GPU）
+- 准备Ascend或GPU处理器搭建硬件环境。
+- 框架
+- [MindSpore](https://www.mindspore.cn/install)
+- 如需查看详情，请参见如下资源：
+- [MindSpore教程](https://www.mindspore.cn/tutorials/zh-CN/master/index.html)
+- [MindSpore Python API](https://www.mindspore.cn/docs/api/zh-CN/master/index.html)
+
+# 脚本说明
+
+## 脚本及样例代码
+
+```shell
+├── model_zoo
+ ├── README.md // 所有模型的说明
+ ├── densenet
+ ├── README_CN.md // DenseNet相关说明
+ ├── ascend310_infer // 实现310推理源代码
+ ├── scripts
+ │ ├── run_infer_310.sh // Ascend 310 推理shell脚本
+ ├── src
+ │ ├── densenet121.py // densenet-121模型文件
+ │ ├── densenet161.py // densenet-161模型文件
+ │ ├── densenet169.py // densenet-169模型文件
+ │ ├── densenet201.py // densenet-201模型文件
+ ├── export.py // 导出脚本
+ ├── postprocess.py // 310 推理后处理脚本
+```
+
+## 导出过程
+
+### 导出
+
+```shell
+python export.py --backbone [NET_NAME] --ckpt_path [CKPT_PATH] --device_target [DEVICE_TARGET] --device_id 0 --file_format [EXPORT_FORMAT] --file_name [FILE_NAME]
+```
+
+`backbone` 可选 ["densenet121", "densenet161", "densenet169", "densenet201"]
+`EXPORT_FORMAT` 可选 ["AIR", "MINDIR"]
+
+## 推理过程
+
+### 推理
+
+在推理之前需要先导出模型，AIR模型只能在昇腾910环境上导出，MINDIR可以在任意环境上导出。
+
+```shell
+# 昇腾310 推理
+bash run_infer_310.sh [MINDIR_PATH] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+```
+
+-注: Densnet系列网络使用ImageNet数据集,图片的label是将文件夹排序后从0开始编号所得的数字.
+
+推理的结果保存在当前目录下，在acc.log日志文件中可以找到类似以下的结果。
+Densenet-121网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=37244, tot=50000, acc=74.49%
+ after allreduce eval: top5_correct=46060, tot=50000, acc=92.12%
+ ```
+
+Densenet-161网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=38773, tot=50000, acc=77.55%
+ after allreduce eval: top5_correct=46971, tot=50000, acc=93.94%
+ ```
+
+Densenet-169网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=38076, tot=50000, acc=76.15%
+ after allreduce eval: top5_correct=46563, tot=50000, acc=93.13%
+ ```
+
+Densenet-201网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=38469, tot=50000, acc=76.94%
+ after allreduce eval: top5_correct=46776, tot=50000, acc=93.55%
+ ```
+
+# ModelZoo主页
+
+ 请浏览官网[主页](https://gitee.com/mindspore/models)。
diff --git a/research/cvtmodel/densenet/ascend310_infer/CMakeLists.txt b/research/cvtmodel/densenet/ascend310_infer/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..ee3c85447340e0449ff2b70ed24f60a17e07b2b6
--- /dev/null
+++ b/research/cvtmodel/densenet/ascend310_infer/CMakeLists.txt
@@ -0,0 +1,14 @@
+cmake_minimum_required(VERSION 3.14.1)
+project(Ascend310Infer)
+add_compile_definitions(_GLIBCXX_USE_CXX11_ABI=0)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O0 -g -std=c++17 -Werror -Wall -fPIE -Wl,--allow-shlib-undefined")
+set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
+option(MINDSPORE_PATH "mindspore install path" "")
+include_directories(${MINDSPORE_PATH})
+include_directories(${MINDSPORE_PATH}/include)
+include_directories(${PROJECT_SRC_ROOT})
+find_library(MS_LIB libmindspore.so ${MINDSPORE_PATH}/lib)
+file(GLOB_RECURSE MD_LIB ${MINDSPORE_PATH}/_c_dataengine*)
+
+add_executable(main src/main.cc src/utils.cc)
+target_link_libraries(main ${MS_LIB} ${MD_LIB} gflags)
diff --git a/research/cvtmodel/densenet/ascend310_infer/build.sh b/research/cvtmodel/densenet/ascend310_infer/build.sh
new file mode 100644
index 0000000000000000000000000000000000000000..770a8851efade7f352039fc8665d307ae1abbb00
--- /dev/null
+++ b/research/cvtmodel/densenet/ascend310_infer/build.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [ ! -d out ]; then
+ mkdir out
+fi
+cd out || exit
+cmake .. \
+ -DMINDSPORE_PATH="`pip show mindspore-ascend | grep Location | awk '{print $2"/mindspore"}' | xargs realpath`"
+make
diff --git a/research/cvtmodel/densenet/ascend310_infer/inc/utils.h b/research/cvtmodel/densenet/ascend310_infer/inc/utils.h
new file mode 100644
index 0000000000000000000000000000000000000000..efebe03a8c1179f5a1f9d5f7ee07e0352a9937c6
--- /dev/null
+++ b/research/cvtmodel/densenet/ascend310_infer/inc/utils.h
@@ -0,0 +1,32 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef MINDSPORE_INFERENCE_UTILS_H_
+#define MINDSPORE_INFERENCE_UTILS_H_
+
+#include <sys/stat.h>
+#include <dirent.h>
+#include <vector>
+#include <string>
+#include <memory>
+#include "include/api/types.h"
+
+std::vector<std::string> GetAllFiles(std::string_view dirName);
+DIR *OpenDir(std::string_view dirName);
+std::string RealPath(std::string_view path);
+mindspore::MSTensor ReadFileToTensor(const std::string &file);
+int WriteResult(const std::string& imageFile, const std::vector<mindspore::MSTensor> &outputs);
+#endif
diff --git a/research/cvtmodel/densenet/ascend310_infer/src/main.cc b/research/cvtmodel/densenet/ascend310_infer/src/main.cc
new file mode 100644
index 0000000000000000000000000000000000000000..a3885312e2690acf34137bfd4490dd0401c21de4
--- /dev/null
+++ b/research/cvtmodel/densenet/ascend310_infer/src/main.cc
@@ -0,0 +1,166 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <sys/time.h>
+#include <gflags/gflags.h>
+#include <dirent.h>
+#include <iostream>
+#include <string>
+#include <algorithm>
+#include <iosfwd>
+#include <vector>
+#include <fstream>
+
+#include "../inc/utils.h"
+#include "include/dataset/execute.h"
+#include "include/dataset/transforms.h"
+#include "include/dataset/vision.h"
+#include "include/dataset/vision_ascend.h"
+#include "include/api/types.h"
+#include "include/api/model.h"
+#include "include/api/serialization.h"
+#include "include/api/context.h"
+
+using mindspore::Serialization;
+using mindspore::Model;
+using mindspore::Context;
+using mindspore::Status;
+using mindspore::ModelType;
+using mindspore::Graph;
+using mindspore::GraphCell;
+using mindspore::kSuccess;
+using mindspore::MSTensor;
+using mindspore::DataType;
+using mindspore::dataset::Execute;
+using mindspore::dataset::TensorTransform;
+using mindspore::dataset::vision::Decode;
+using mindspore::dataset::vision::Resize;
+using mindspore::dataset::vision::CenterCrop;
+using mindspore::dataset::vision::Normalize;
+using mindspore::dataset::vision::HWC2CHW;
+
+using mindspore::dataset::transforms::TypeCast;
+
+
+DEFINE_string(model_path, "", "model path");
+DEFINE_string(dataset, "ImageNet", "dataset: ImageNet or Cifar10");
+DEFINE_string(dataset_path, ".", "dataset path");
+DEFINE_int32(device_id, 0, "device id");
+
+int main(int argc, char **argv) {
+ gflags::ParseCommandLineFlags(&argc, &argv, true);
+ if (RealPath(FLAGS_model_path).empty()) {
+ std::cout << "Invalid model" << std::endl;
+ return 1;
+ }
+
+ std::transform(FLAGS_dataset.begin(), FLAGS_dataset.end(), FLAGS_dataset.begin(), ::tolower);
+
+ auto context = std::make_shared<Context>();
+ auto ascend310_info = std::make_shared<mindspore::Ascend310DeviceInfo>();
+ ascend310_info->SetDeviceID(FLAGS_device_id);
+ context->MutableDeviceInfo().push_back(ascend310_info);
+
+ Graph graph;
+ Status ret = Serialization::Load(FLAGS_model_path, ModelType::kMindIR, &graph);
+ if (ret != kSuccess) {
+ std::cout << "Load model failed." << std::endl;
+ return 1;
+ }
+
+ Model model;
+ ret = model.Build(GraphCell(graph), context);
+ if (ret != kSuccess) {
+ std::cout << "ERROR: Build failed." << std::endl;
+ return 1;
+ }
+
+ std::vector<MSTensor> modelInputs = model.GetInputs();
+
+ auto all_files = GetAllFiles(FLAGS_dataset_path);
+ if (all_files.empty()) {
+ std::cout << "ERROR: no input data." << std::endl;
+ return 1;
+ }
+
+ std::shared_ptr<TensorTransform> decode(new Decode());
+ std::shared_ptr<TensorTransform> resize(new Resize({256, 256}));
+ std::shared_ptr<TensorTransform> centerCrop(new CenterCrop({224, 224}));
+ std::shared_ptr<TensorTransform> normImageNet(new Normalize({123.675, 116.28, 103.53}, {58.395, 57.12, 57.375}));
+ std::shared_ptr<TensorTransform> normCifar10(new Normalize({127.5, 127.5, 127.5}, {127.5, 127.5, 127.5}));
+ std::shared_ptr<TensorTransform> hwc2chw(new HWC2CHW());
+
+ mindspore::dataset::Execute transformImageNet({decode, resize, centerCrop, normImageNet, hwc2chw});
+ mindspore::dataset::Execute transformCifar10({decode, normCifar10, hwc2chw});
+
+ std::map<double, double> costTime_map;
+
+ size_t size = all_files.size();
+ for (size_t i = 0; i < size; ++i) {
+ struct timeval start;
+ struct timeval end;
+ double startTime_ms;
+ double endTime_ms;
+ std::vector<MSTensor> inputs;
+ std::vector<MSTensor> outputs;
+
+ std::cout << "Start predict input files:" << all_files[i] << std::endl;
+ mindspore::MSTensor image = ReadFileToTensor(all_files[i]);
+
+ if (FLAGS_dataset.compare("cifar10") == 0) {
+ transformCifar10(image, &image);
+ } else if (FLAGS_dataset.compare("imagenet") == 0) {
+ transformImageNet(image, &image);
+ } else {
+ std::cout << "unsupported dataset ...";
+ return 1;
+ }
+
+ inputs.emplace_back(modelInputs[0].Name(), modelInputs[0].DataType(), modelInputs[0].Shape(),
+ image.Data().get(), image.DataSize());
+
+ gettimeofday(&start, NULL);
+ model.Predict(inputs, &outputs);
+ gettimeofday(&end, NULL);
+
+ startTime_ms = (1.0 * start.tv_sec * 1000000 + start.tv_usec) / 1000;
+ endTime_ms = (1.0 * end.tv_sec * 1000000 + end.tv_usec) / 1000;
+ costTime_map.insert(std::pair<double, double>(startTime_ms, endTime_ms));
+ int rst = WriteResult(all_files[i], outputs);
+ if (rst != 0) {
+ std::cout << "write result failed." << std::endl;
+ return rst;
+ }
+ }
+ double average = 0.0;
+ int infer_cnt = 0;
+ char tmpCh[256] = {0};
+ for (auto iter = costTime_map.begin(); iter != costTime_map.end(); iter++) {
+ double diff = 0.0;
+ diff = iter->second - iter->first;
+ average += diff;
+ infer_cnt++;
+ }
+
+ average = average/infer_cnt;
+ std::cout << "NN inference cost average time: "<< average << "ms of infer_count " << infer_cnt << std::endl;
+ std::string file_name = "./time_Result" + std::string("/test_perform_static.txt");
+ std::ofstream file_stream(file_name.c_str(), std::ios::trunc);
+ file_stream << tmpCh;
+ file_stream.close();
+ costTime_map.clear();
+ return 0;
+}
diff --git a/research/cvtmodel/densenet/ascend310_infer/src/utils.cc b/research/cvtmodel/densenet/ascend310_infer/src/utils.cc
new file mode 100644
index 0000000000000000000000000000000000000000..653b1de44962614ac77d44481f8d1a7bde52caaf
--- /dev/null
+++ b/research/cvtmodel/densenet/ascend310_infer/src/utils.cc
@@ -0,0 +1,160 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "inc/utils.h"
+
+#include <fstream>
+#include <algorithm>
+#include <iostream>
+
+using mindspore::MSTensor;
+using mindspore::DataType;
+
+std::vector<std::string> GetAllFiles(std::string_view dirName) {
+ struct dirent *filename;
+ DIR *dir = OpenDir(dirName);
+ if (dir == nullptr) {
+ return {};
+ }
+ std::vector<std::string> dirs;
+ std::vector<std::string> files;
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == "..") {
+ continue;
+ } else if (filename->d_type == DT_DIR) {
+ dirs.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else if (filename->d_type == DT_REG) {
+ files.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else {
+ continue;
+ }
+ }
+
+ for (auto d : dirs) {
+ dir = OpenDir(d);
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == ".." || filename->d_type != DT_REG) {
+ continue;
+ }
+ files.emplace_back(std::string(d) + "/" + filename->d_name);
+ }
+ }
+ std::sort(files.begin(), files.end());
+ for (auto &f : files) {
+ std::cout << "image file: " << f << std::endl;
+ }
+ return files;
+}
+
+int WriteResult(const std::string& imageFile, const std::vector<MSTensor> &outputs) {
+ std::string homePath = "./result_Files";
+ const int INVALID_POINTER = -1;
+ const int ERROR = -2;
+ for (size_t i = 0; i < outputs.size(); ++i) {
+ size_t outputSize;
+ std::shared_ptr<const void> netOutput;
+ netOutput = outputs[i].Data();
+ outputSize = outputs[i].DataSize();
+ int pos = imageFile.rfind('/');
+ std::string fileName(imageFile, pos + 1);
+ fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), '_' + std::to_string(i) + ".bin");
+ std::string outFileName = homePath + "/" + fileName;
+ FILE * outputFile = fopen(outFileName.c_str(), "wb");
+ if (outputFile == nullptr) {
+ std::cout << "open result file " << outFileName << " failed" << std::endl;
+ return INVALID_POINTER;
+ }
+ size_t size = fwrite(netOutput.get(), sizeof(char), outputSize, outputFile);
+ if (size != outputSize) {
+ fclose(outputFile);
+ outputFile = nullptr;
+ std::cout << "write result file " << outFileName << " failed, write size[" << size <<
+ "] is smaller than output size[" << outputSize << "], maybe the disk is full." << std::endl;
+ return ERROR;
+ }
+ fclose(outputFile);
+ outputFile = nullptr;
+ }
+ return 0;
+}
+
+mindspore::MSTensor ReadFileToTensor(const std::string &file) {
+ if (file.empty()) {
+ std::cout << "Pointer file is nullptr" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ std::ifstream ifs(file);
+ if (!ifs.good()) {
+ std::cout << "File: " << file << " is not exist" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ if (!ifs.is_open()) {
+ std::cout << "File: " << file << "open failed" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ ifs.seekg(0, std::ios::end);
+ size_t size = ifs.tellg();
+ mindspore::MSTensor buffer(file, mindspore::DataType::kNumberTypeUInt8, {static_cast<int64_t>(size)}, nullptr, size);
+
+ ifs.seekg(0, std::ios::beg);
+ ifs.read(reinterpret_cast<char *>(buffer.MutableData()), size);
+ ifs.close();
+
+ return buffer;
+}
+
+
+DIR *OpenDir(std::string_view dirName) {
+ if (dirName.empty()) {
+ std::cout << " dirName is null ! " << std::endl;
+ return nullptr;
+ }
+ std::string realPath = RealPath(dirName);
+ struct stat s;
+ lstat(realPath.c_str(), &s);
+ if (!S_ISDIR(s.st_mode)) {
+ std::cout << "dirName is not a valid directory !" << std::endl;
+ return nullptr;
+ }
+ DIR *dir;
+ dir = opendir(realPath.c_str());
+ if (dir == nullptr) {
+ std::cout << "Can not open dir " << dirName << std::endl;
+ return nullptr;
+ }
+ std::cout << "Successfully opened the dir " << dirName << std::endl;
+ return dir;
+}
+
+std::string RealPath(std::string_view path) {
+ char realPathMem[PATH_MAX] = {0};
+ char *realPathRet = nullptr;
+ realPathRet = realpath(path.data(), realPathMem);
+
+ if (realPathRet == nullptr) {
+ std::cout << "File: " << path << " is not exist.";
+ return "";
+ }
+
+ std::string realPath(realPathMem);
+ std::cout << path << " realpath is: " << realPath << std::endl;
+ return realPath;
+}
diff --git a/research/cvtmodel/densenet/export.py b/research/cvtmodel/densenet/export.py
new file mode 100644
index 0000000000000000000000000000000000000000..81f02f491feee0ca045a24a7e71d3c8b5d2f8597
--- /dev/null
+++ b/research/cvtmodel/densenet/export.py
@@ -0,0 +1,59 @@
+# Copyright 2020-2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+"""export checkpoint file into air, onnx, mindir models
+ Suggest run as python export.py --file_name [file_name] --ckpt_files [ckpt path] --file_format [file format]
+"""
+import argparse
+import numpy as np
+import mindspore as ms
+from mindspore import context, Tensor
+from mindspore.train.serialization import export, load_checkpoint, load_param_into_net
+
+parser = argparse.ArgumentParser(description='post process for 310 inference')
+parser.add_argument("--backbone", type=str, required=True, default="densenet161", help="model backbone")
+parser.add_argument("--ckpt_path", type=str, required=True, help="checkpoint file path")
+parser.add_argument("--file_name", type=str, default="densenet161", help="file name")
+parser.add_argument("--file_format", type=str, default="MINDIR", choices=["MINDIR", "AIR"], help="file format")
+parser.add_argument("--device_target", type=str, default="Ascend", choices=["Ascend", "GPU"], help="device target")
+parser.add_argument("--device_id", type=int, default=0, help="device target")
+args = parser.parse_args()
+
+context.set_context(mode=context.GRAPH_MODE, device_target=args.device_target)
+
+def model_export():
+ '''main export func'''
+ if args.device_target == "Ascend":
+ context.set_context(device_id=args.device_id)
+
+ if args.backbone == "densenet121":
+ from src.densenet121 import MainModel
+ elif args.backbone == "densenet161":
+ from src.densenet161 import MainModel
+ elif args.backbone == "densenet169":
+ from src.densenet169 import MainModel
+ elif args.backbone == "densenet201":
+ from src.densenet201 import MainModel
+
+ net = MainModel()
+
+ param_dict = load_checkpoint(args.ckpt_path)
+ load_param_into_net(net, param_dict)
+
+ input_arr = Tensor(np.zeros([1, 3, 224, 224]), ms.float32)
+ export(net, input_arr, file_name=args.file_name, file_format=args.file_format)
+
+if __name__ == '__main__':
+ model_export()
diff --git a/research/cvtmodel/densenet/postprocess.py b/research/cvtmodel/densenet/postprocess.py
new file mode 100644
index 0000000000000000000000000000000000000000..ee41ac3961e6054ca9e17bcd3ab8cb210ffa9e84
--- /dev/null
+++ b/research/cvtmodel/densenet/postprocess.py
@@ -0,0 +1,82 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+'''post process for 310 inference'''
+import os
+import argparse
+import numpy as np
+
+parser = argparse.ArgumentParser(description='post process for 310 inference')
+parser.add_argument("--dataset", type=str, default="imagenet", help="result file path")
+parser.add_argument("--result_path", type=str, required=True, help="result file path")
+parser.add_argument("--label_file", type=str, required=True, help="label file")
+args = parser.parse_args()
+
+def get_top5_acc(top5_arg, gt_class):
+ sub_count = 0
+ for top5, gt in zip(top5_arg, gt_class):
+ if gt in top5:
+ sub_count += 1
+ return sub_count
+
+def read_label(label_file):
+ '''read label file'''
+ f = open(label_file, "r")
+ lines = f.readlines()
+
+ img_label = {}
+ for line in lines:
+ img_id = line.split(":")[0]
+ label = line.split(":")[1]
+ img_label[img_id] = label
+
+ return img_label
+
+def cal_acc(dataset, result_path, label_file):
+ '''main acc calculation function'''
+ img_label = read_label(label_file)
+
+ img_tot = 0
+ top1_correct = 0
+ top5_correct = 0
+
+ files = os.listdir(result_path)
+ for file in files:
+ full_file_path = os.path.join(result_path, file)
+ if os.path.isfile(full_file_path):
+ result = np.fromfile(full_file_path, dtype=np.float32).reshape(1, 1000)
+ gt_classes = int(img_label[file[:-6]])
+
+ top1_output = np.argmax(result, (-1))
+ top5_output = np.argsort(result)[:, -5:]
+
+ t1_correct = np.equal(top1_output, gt_classes).sum()
+ top1_correct += t1_correct
+ top5_correct += get_top5_acc(top5_output, [gt_classes])
+ img_tot += 1
+
+ results = [[top1_correct], [top5_correct], [img_tot]]
+
+ results = np.array(results)
+ top1_correct = results[0, 0]
+ top5_correct = results[1, 0]
+ img_tot = results[2, 0]
+ acc1 = 100.0 * top1_correct / img_tot
+ acc5 = 100.0 * top5_correct / img_tot
+ print('after allreduce eval: top1_correct={}, tot={}, acc={:.2f}%'.format(top1_correct, img_tot, acc1))
+ if dataset == 'imagenet':
+ print('after allreduce eval: top5_correct={}, tot={}, acc={:.2f}%'.format(top5_correct, img_tot, acc5))
+
+if __name__ == "__main__":
+ cal_acc(args.dataset, args.result_path, args.label_file)
diff --git a/research/cvtmodel/densenet/scripts/run_infer_310.sh b/research/cvtmodel/densenet/scripts/run_infer_310.sh
new file mode 100644
index 0000000000000000000000000000000000000000..e4c119abc74e0ac0e0d699e638bef77579137b46
--- /dev/null
+++ b/research/cvtmodel/densenet/scripts/run_infer_310.sh
@@ -0,0 +1,106 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [[ $# -lt 4 || $# -gt 5 ]]; then
+ echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+ DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
+exit 1
+fi
+
+get_real_path(){
+ if [ "${1:0:1}" == "/" ]; then
+ echo "$1"
+ else
+ echo "$(realpath -m $PWD/$1)"
+ fi
+}
+
+model=$(get_real_path $1)
+dataset=$2
+data_path=$(get_real_path $3)
+label_file=$(get_real_path $4)
+
+device_id=0
+
+if [ $# == 4 ]; then
+ device_id=$4
+fi
+
+echo $model
+echo $dataset
+echo $data_path
+echo $label_file
+echo $device_id
+
+export ASCEND_HOME=/usr/local/Ascend/
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+ export PATH=$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export TBE_IMPL_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+ export PYTHONPATH=${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+ export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi
+
+function compile_app()
+{
+ cd ../ascend310_infer || exit
+ if [ -f "Makefile" ]; then
+ make clean
+ fi
+ sh build.sh &> build.log
+
+ if [ $? -ne 0 ]; then
+ echo "compile app code failed"
+ exit 1
+ fi
+ cd - || exit
+}
+
+function infer()
+{
+ if [ -d result_Files ]; then
+ rm -rf ./result_Files
+ fi
+ if [ -d time_Result ]; then
+ rm -rf ./time_Result
+ fi
+ mkdir result_Files
+ mkdir time_Result
+ ../ascend310_infer/out/main --model_path=$model --dataset=$dataset --dataset_path=$data_path --device_id=$device_id &> infer.log
+
+ if [ $? -ne 0 ]; then
+ echo "execute inference failed"
+ exit 1
+ fi
+}
+
+function cal_acc()
+{
+ python ../postprocess.py --dataset=$dataset --label_file=$label_file --result_path=result_Files &> acc.log
+ if [ $? -ne 0 ]; then
+ echo "calculate accuracy failed"
+ exit 1
+ fi
+}
+
+compile_app
+infer
+cal_acc
diff --git a/research/cvtmodel/densenet/src/densenet121.py b/research/cvtmodel/densenet/src/densenet121.py
new file mode 100644
index 0000000000000000000000000000000000000000..deea310222f87de786a59925c076070be945b249
--- /dev/null
+++ b/research/cvtmodel/densenet/src/densenet121.py
@@ -0,0 +1,541 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module0(nn.Cell):
+ def __init__(self, batchnorm2d_1_num_features, conv2d_3_in_channels):
+ super(Module0, self).__init__()
+ self.concat_0 = P.Concat(axis=1)
+ self.batchnorm2d_1 = nn.BatchNorm2d(num_features=batchnorm2d_1_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_4 = nn.ReLU()
+ self.conv2d_5 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ opt_concat_0 = self.concat_0((x,))
+ opt_batchnorm2d_1 = self.batchnorm2d_1(opt_concat_0)
+ opt_relu_2 = self.relu_2(opt_batchnorm2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_relu_4 = self.relu_4(opt_conv2d_3)
+ opt_conv2d_5 = self.conv2d_5(opt_relu_4)
+ return opt_conv2d_5
+
+
+class Module1(nn.Cell):
+ def __init__(self, batchnorm2d_0_num_features):
+ super(Module1, self).__init__()
+ self.batchnorm2d_0 = nn.BatchNorm2d(num_features=batchnorm2d_0_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_batchnorm2d_0 = self.batchnorm2d_0(x)
+ opt_relu_1 = self.relu_1(opt_batchnorm2d_0)
+ return opt_relu_1
+
+
+class Module4(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, module1_0_batchnorm2d_0_num_features):
+ super(Module4, self).__init__()
+ self.module1_0 = Module1(batchnorm2d_0_num_features=module1_0_batchnorm2d_0_num_features)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.conv2d_2 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ module1_0_opt = self.module1_0(x)
+ opt_conv2d_0 = self.conv2d_0(module1_0_opt)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_conv2d_2 = self.conv2d_2(opt_relu_1)
+ return opt_conv2d_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(7, 7),
+ stride=(2, 2),
+ padding=(3, 3, 3, 3),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.pad_maxpool2d_2 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_2 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module0_0 = Module0(batchnorm2d_1_num_features=64, conv2d_3_in_channels=64)
+ self.concat_9 = P.Concat(axis=1)
+ self.module4_0 = Module4(conv2d_0_in_channels=96, module1_0_batchnorm2d_0_num_features=96)
+ self.concat_15 = P.Concat(axis=1)
+ self.module4_1 = Module4(conv2d_0_in_channels=128, module1_0_batchnorm2d_0_num_features=128)
+ self.concat_21 = P.Concat(axis=1)
+ self.module4_2 = Module4(conv2d_0_in_channels=160, module1_0_batchnorm2d_0_num_features=160)
+ self.concat_27 = P.Concat(axis=1)
+ self.module4_3 = Module4(conv2d_0_in_channels=192, module1_0_batchnorm2d_0_num_features=192)
+ self.concat_33 = P.Concat(axis=1)
+ self.module4_4 = Module4(conv2d_0_in_channels=224, module1_0_batchnorm2d_0_num_features=224)
+ self.concat_39 = P.Concat(axis=1)
+ self.module1_0 = Module1(batchnorm2d_0_num_features=256)
+ self.conv2d_42 = nn.Conv2d(in_channels=256,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_43 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_43 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_1 = Module0(batchnorm2d_1_num_features=128, conv2d_3_in_channels=128)
+ self.concat_50 = P.Concat(axis=1)
+ self.module4_5 = Module4(conv2d_0_in_channels=160, module1_0_batchnorm2d_0_num_features=160)
+ self.concat_56 = P.Concat(axis=1)
+ self.module4_6 = Module4(conv2d_0_in_channels=192, module1_0_batchnorm2d_0_num_features=192)
+ self.concat_62 = P.Concat(axis=1)
+ self.module4_7 = Module4(conv2d_0_in_channels=224, module1_0_batchnorm2d_0_num_features=224)
+ self.concat_68 = P.Concat(axis=1)
+ self.module4_8 = Module4(conv2d_0_in_channels=256, module1_0_batchnorm2d_0_num_features=256)
+ self.concat_74 = P.Concat(axis=1)
+ self.module4_9 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_80 = P.Concat(axis=1)
+ self.module4_10 = Module4(conv2d_0_in_channels=320, module1_0_batchnorm2d_0_num_features=320)
+ self.concat_86 = P.Concat(axis=1)
+ self.module4_11 = Module4(conv2d_0_in_channels=352, module1_0_batchnorm2d_0_num_features=352)
+ self.concat_92 = P.Concat(axis=1)
+ self.module4_12 = Module4(conv2d_0_in_channels=384, module1_0_batchnorm2d_0_num_features=384)
+ self.concat_98 = P.Concat(axis=1)
+ self.module4_13 = Module4(conv2d_0_in_channels=416, module1_0_batchnorm2d_0_num_features=416)
+ self.concat_104 = P.Concat(axis=1)
+ self.module4_14 = Module4(conv2d_0_in_channels=448, module1_0_batchnorm2d_0_num_features=448)
+ self.concat_110 = P.Concat(axis=1)
+ self.module4_15 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_116 = P.Concat(axis=1)
+ self.module1_1 = Module1(batchnorm2d_0_num_features=512)
+ self.conv2d_119 = nn.Conv2d(in_channels=512,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_120 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_120 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_2 = Module0(batchnorm2d_1_num_features=256, conv2d_3_in_channels=256)
+ self.concat_127 = P.Concat(axis=1)
+ self.module4_16 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_133 = P.Concat(axis=1)
+ self.module4_17 = Module4(conv2d_0_in_channels=320, module1_0_batchnorm2d_0_num_features=320)
+ self.concat_139 = P.Concat(axis=1)
+ self.module4_18 = Module4(conv2d_0_in_channels=352, module1_0_batchnorm2d_0_num_features=352)
+ self.concat_145 = P.Concat(axis=1)
+ self.module4_19 = Module4(conv2d_0_in_channels=384, module1_0_batchnorm2d_0_num_features=384)
+ self.concat_151 = P.Concat(axis=1)
+ self.module4_20 = Module4(conv2d_0_in_channels=416, module1_0_batchnorm2d_0_num_features=416)
+ self.concat_157 = P.Concat(axis=1)
+ self.module4_21 = Module4(conv2d_0_in_channels=448, module1_0_batchnorm2d_0_num_features=448)
+ self.concat_163 = P.Concat(axis=1)
+ self.module4_22 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_169 = P.Concat(axis=1)
+ self.module4_23 = Module4(conv2d_0_in_channels=512, module1_0_batchnorm2d_0_num_features=512)
+ self.concat_175 = P.Concat(axis=1)
+ self.module4_24 = Module4(conv2d_0_in_channels=544, module1_0_batchnorm2d_0_num_features=544)
+ self.concat_181 = P.Concat(axis=1)
+ self.module4_25 = Module4(conv2d_0_in_channels=576, module1_0_batchnorm2d_0_num_features=576)
+ self.concat_187 = P.Concat(axis=1)
+ self.module4_26 = Module4(conv2d_0_in_channels=608, module1_0_batchnorm2d_0_num_features=608)
+ self.concat_193 = P.Concat(axis=1)
+ self.module4_27 = Module4(conv2d_0_in_channels=640, module1_0_batchnorm2d_0_num_features=640)
+ self.concat_199 = P.Concat(axis=1)
+ self.module4_28 = Module4(conv2d_0_in_channels=672, module1_0_batchnorm2d_0_num_features=672)
+ self.concat_205 = P.Concat(axis=1)
+ self.module4_29 = Module4(conv2d_0_in_channels=704, module1_0_batchnorm2d_0_num_features=704)
+ self.concat_211 = P.Concat(axis=1)
+ self.module4_30 = Module4(conv2d_0_in_channels=736, module1_0_batchnorm2d_0_num_features=736)
+ self.concat_217 = P.Concat(axis=1)
+ self.module4_31 = Module4(conv2d_0_in_channels=768, module1_0_batchnorm2d_0_num_features=768)
+ self.concat_223 = P.Concat(axis=1)
+ self.module4_32 = Module4(conv2d_0_in_channels=800, module1_0_batchnorm2d_0_num_features=800)
+ self.concat_229 = P.Concat(axis=1)
+ self.module4_33 = Module4(conv2d_0_in_channels=832, module1_0_batchnorm2d_0_num_features=832)
+ self.concat_235 = P.Concat(axis=1)
+ self.module4_34 = Module4(conv2d_0_in_channels=864, module1_0_batchnorm2d_0_num_features=864)
+ self.concat_241 = P.Concat(axis=1)
+ self.module4_35 = Module4(conv2d_0_in_channels=896, module1_0_batchnorm2d_0_num_features=896)
+ self.concat_247 = P.Concat(axis=1)
+ self.module4_36 = Module4(conv2d_0_in_channels=928, module1_0_batchnorm2d_0_num_features=928)
+ self.concat_253 = P.Concat(axis=1)
+ self.module4_37 = Module4(conv2d_0_in_channels=960, module1_0_batchnorm2d_0_num_features=960)
+ self.concat_259 = P.Concat(axis=1)
+ self.module4_38 = Module4(conv2d_0_in_channels=992, module1_0_batchnorm2d_0_num_features=992)
+ self.concat_265 = P.Concat(axis=1)
+ self.module1_2 = Module1(batchnorm2d_0_num_features=1024)
+ self.conv2d_268 = nn.Conv2d(in_channels=1024,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_269 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_269 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_3 = Module0(batchnorm2d_1_num_features=512, conv2d_3_in_channels=512)
+ self.concat_276 = P.Concat(axis=1)
+ self.module4_39 = Module4(conv2d_0_in_channels=544, module1_0_batchnorm2d_0_num_features=544)
+ self.concat_282 = P.Concat(axis=1)
+ self.module4_40 = Module4(conv2d_0_in_channels=576, module1_0_batchnorm2d_0_num_features=576)
+ self.concat_288 = P.Concat(axis=1)
+ self.module4_41 = Module4(conv2d_0_in_channels=608, module1_0_batchnorm2d_0_num_features=608)
+ self.concat_294 = P.Concat(axis=1)
+ self.module4_42 = Module4(conv2d_0_in_channels=640, module1_0_batchnorm2d_0_num_features=640)
+ self.concat_300 = P.Concat(axis=1)
+ self.module4_43 = Module4(conv2d_0_in_channels=672, module1_0_batchnorm2d_0_num_features=672)
+ self.concat_306 = P.Concat(axis=1)
+ self.module4_44 = Module4(conv2d_0_in_channels=704, module1_0_batchnorm2d_0_num_features=704)
+ self.concat_312 = P.Concat(axis=1)
+ self.module4_45 = Module4(conv2d_0_in_channels=736, module1_0_batchnorm2d_0_num_features=736)
+ self.concat_318 = P.Concat(axis=1)
+ self.module4_46 = Module4(conv2d_0_in_channels=768, module1_0_batchnorm2d_0_num_features=768)
+ self.concat_324 = P.Concat(axis=1)
+ self.module4_47 = Module4(conv2d_0_in_channels=800, module1_0_batchnorm2d_0_num_features=800)
+ self.concat_330 = P.Concat(axis=1)
+ self.module4_48 = Module4(conv2d_0_in_channels=832, module1_0_batchnorm2d_0_num_features=832)
+ self.concat_336 = P.Concat(axis=1)
+ self.module4_49 = Module4(conv2d_0_in_channels=864, module1_0_batchnorm2d_0_num_features=864)
+ self.concat_342 = P.Concat(axis=1)
+ self.module4_50 = Module4(conv2d_0_in_channels=896, module1_0_batchnorm2d_0_num_features=896)
+ self.concat_348 = P.Concat(axis=1)
+ self.module4_51 = Module4(conv2d_0_in_channels=928, module1_0_batchnorm2d_0_num_features=928)
+ self.concat_354 = P.Concat(axis=1)
+ self.module4_52 = Module4(conv2d_0_in_channels=960, module1_0_batchnorm2d_0_num_features=960)
+ self.concat_360 = P.Concat(axis=1)
+ self.module4_53 = Module4(conv2d_0_in_channels=992, module1_0_batchnorm2d_0_num_features=992)
+ self.concat_366 = P.Concat(axis=1)
+ self.module1_3 = Module1(batchnorm2d_0_num_features=1024)
+ self.avgpool2d_369 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.flatten_370 = nn.Flatten()
+ self.dense_371 = nn.Dense(in_channels=1024, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_maxpool2d_2 = self.pad_maxpool2d_2(opt_relu_1)
+ opt_maxpool2d_2 = self.maxpool2d_2(opt_maxpool2d_2)
+ module0_0_opt = self.module0_0(opt_maxpool2d_2)
+ opt_concat_9 = self.concat_9((opt_maxpool2d_2, module0_0_opt,))
+ module4_0_opt = self.module4_0(opt_concat_9)
+ opt_concat_15 = self.concat_15((opt_maxpool2d_2, module0_0_opt, module4_0_opt,))
+ module4_1_opt = self.module4_1(opt_concat_15)
+ opt_concat_21 = self.concat_21((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt,))
+ module4_2_opt = self.module4_2(opt_concat_21)
+ opt_concat_27 = self.concat_27((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt,))
+ module4_3_opt = self.module4_3(opt_concat_27)
+ opt_concat_33 = self.concat_33(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt,
+ ))
+ module4_4_opt = self.module4_4(opt_concat_33)
+ opt_concat_39 = self.concat_39(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt, module4_4_opt,
+ ))
+ module1_0_opt = self.module1_0(opt_concat_39)
+ opt_conv2d_42 = self.conv2d_42(module1_0_opt)
+ opt_avgpool2d_43 = self.pad_avgpool2d_43(opt_conv2d_42)
+ opt_avgpool2d_43 = self.avgpool2d_43(opt_avgpool2d_43)
+ module0_1_opt = self.module0_1(opt_avgpool2d_43)
+ opt_concat_50 = self.concat_50((opt_avgpool2d_43, module0_1_opt,))
+ module4_5_opt = self.module4_5(opt_concat_50)
+ opt_concat_56 = self.concat_56((opt_avgpool2d_43, module0_1_opt, module4_5_opt,))
+ module4_6_opt = self.module4_6(opt_concat_56)
+ opt_concat_62 = self.concat_62((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt,))
+ module4_7_opt = self.module4_7(opt_concat_62)
+ opt_concat_68 = self.concat_68((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,))
+ module4_8_opt = self.module4_8(opt_concat_68)
+ opt_concat_74 = self.concat_74(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt,
+ ))
+ module4_9_opt = self.module4_9(opt_concat_74)
+ opt_concat_80 = self.concat_80(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ ))
+ module4_10_opt = self.module4_10(opt_concat_80)
+ opt_concat_86 = self.concat_86((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt,
+ ))
+ module4_11_opt = self.module4_11(opt_concat_86)
+ opt_concat_92 = self.concat_92((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt,
+ ))
+ module4_12_opt = self.module4_12(opt_concat_92)
+ opt_concat_98 = self.concat_98((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt, module4_12_opt,
+ ))
+ module4_13_opt = self.module4_13(opt_concat_98)
+ opt_concat_104 = self.concat_104(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt,
+ ))
+ module4_14_opt = self.module4_14(opt_concat_104)
+ opt_concat_110 = self.concat_110(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt,
+ ))
+ module4_15_opt = self.module4_15(opt_concat_110)
+ opt_concat_116 = self.concat_116(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt, module4_15_opt,
+ ))
+ module1_1_opt = self.module1_1(opt_concat_116)
+ opt_conv2d_119 = self.conv2d_119(module1_1_opt)
+ opt_avgpool2d_120 = self.pad_avgpool2d_120(opt_conv2d_119)
+ opt_avgpool2d_120 = self.avgpool2d_120(opt_avgpool2d_120)
+ module0_2_opt = self.module0_2(opt_avgpool2d_120)
+ opt_concat_127 = self.concat_127((opt_avgpool2d_120, module0_2_opt,))
+ module4_16_opt = self.module4_16(opt_concat_127)
+ opt_concat_133 = self.concat_133((opt_avgpool2d_120, module0_2_opt, module4_16_opt,))
+ module4_17_opt = self.module4_17(opt_concat_133)
+ opt_concat_139 = self.concat_139((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,))
+ module4_18_opt = self.module4_18(opt_concat_139)
+ opt_concat_145 = self.concat_145(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt,
+ ))
+ module4_19_opt = self.module4_19(opt_concat_145)
+ opt_concat_151 = self.concat_151(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ ))
+ module4_20_opt = self.module4_20(opt_concat_151)
+ opt_concat_157 = self.concat_157((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt,
+ ))
+ module4_21_opt = self.module4_21(opt_concat_157)
+ opt_concat_163 = self.concat_163((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt, module4_21_opt,
+ ))
+ module4_22_opt = self.module4_22(opt_concat_163)
+ opt_concat_169 = self.concat_169(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt,
+ ))
+ module4_23_opt = self.module4_23(opt_concat_169)
+ opt_concat_175 = self.concat_175(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt,
+ ))
+ module4_24_opt = self.module4_24(opt_concat_175)
+ opt_concat_181 = self.concat_181(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt,
+ ))
+ module4_25_opt = self.module4_25(opt_concat_181)
+ opt_concat_187 = self.concat_187(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ ))
+ module4_26_opt = self.module4_26(opt_concat_187)
+ opt_concat_193 = self.concat_193(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt,
+ ))
+ module4_27_opt = self.module4_27(opt_concat_193)
+ opt_concat_199 = self.concat_199(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt,
+ ))
+ module4_28_opt = self.module4_28(opt_concat_199)
+ opt_concat_205 = self.concat_205(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt,
+ ))
+ module4_29_opt = self.module4_29(opt_concat_205)
+ opt_concat_211 = self.concat_211(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt,
+ ))
+ module4_30_opt = self.module4_30(opt_concat_211)
+ opt_concat_217 = self.concat_217(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt,
+ ))
+ module4_31_opt = self.module4_31(opt_concat_217)
+ opt_concat_223 = self.concat_223(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ ))
+ module4_32_opt = self.module4_32(opt_concat_223)
+ opt_concat_229 = self.concat_229(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt,
+ ))
+ module4_33_opt = self.module4_33(opt_concat_229)
+ opt_concat_235 = self.concat_235(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt,
+ ))
+ module4_34_opt = self.module4_34(opt_concat_235)
+ opt_concat_241 = self.concat_241(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt,
+ ))
+ module4_35_opt = self.module4_35(opt_concat_241)
+ opt_concat_247 = self.concat_247(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt,
+ ))
+ module4_36_opt = self.module4_36(opt_concat_247)
+ opt_concat_253 = self.concat_253(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt,
+ ))
+ module4_37_opt = self.module4_37(opt_concat_253)
+ opt_concat_259 = self.concat_259(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ ))
+ module4_38_opt = self.module4_38(opt_concat_259)
+ opt_concat_265 = self.concat_265(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt,
+ ))
+ module1_2_opt = self.module1_2(opt_concat_265)
+ opt_conv2d_268 = self.conv2d_268(module1_2_opt)
+ opt_avgpool2d_269 = self.pad_avgpool2d_269(opt_conv2d_268)
+ opt_avgpool2d_269 = self.avgpool2d_269(opt_avgpool2d_269)
+ module0_3_opt = self.module0_3(opt_avgpool2d_269)
+ opt_concat_276 = self.concat_276((opt_avgpool2d_269, module0_3_opt,))
+ module4_39_opt = self.module4_39(opt_concat_276)
+ opt_concat_282 = self.concat_282((opt_avgpool2d_269, module0_3_opt, module4_39_opt,))
+ module4_40_opt = self.module4_40(opt_concat_282)
+ opt_concat_288 = self.concat_288((opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt,))
+ module4_41_opt = self.module4_41(opt_concat_288)
+ opt_concat_294 = self.concat_294(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt,
+ ))
+ module4_42_opt = self.module4_42(opt_concat_294)
+ opt_concat_300 = self.concat_300(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ ))
+ module4_43_opt = self.module4_43(opt_concat_300)
+ opt_concat_306 = self.concat_306((opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt,
+ module4_41_opt, module4_42_opt, module4_43_opt,
+ ))
+ module4_44_opt = self.module4_44(opt_concat_306)
+ opt_concat_312 = self.concat_312((opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt,
+ module4_41_opt, module4_42_opt, module4_43_opt, module4_44_opt,
+ ))
+ module4_45_opt = self.module4_45(opt_concat_312)
+ opt_concat_318 = self.concat_318(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt,
+ ))
+ module4_46_opt = self.module4_46(opt_concat_318)
+ opt_concat_324 = self.concat_324(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt,
+ ))
+ module4_47_opt = self.module4_47(opt_concat_324)
+ opt_concat_330 = self.concat_330(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt,
+ ))
+ module4_48_opt = self.module4_48(opt_concat_330)
+ opt_concat_336 = self.concat_336(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ ))
+ module4_49_opt = self.module4_49(opt_concat_336)
+ opt_concat_342 = self.concat_342(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ module4_49_opt,
+ ))
+ module4_50_opt = self.module4_50(opt_concat_342)
+ opt_concat_348 = self.concat_348(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ module4_49_opt, module4_50_opt,
+ ))
+ module4_51_opt = self.module4_51(opt_concat_348)
+ opt_concat_354 = self.concat_354(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ module4_49_opt, module4_50_opt, module4_51_opt,
+ ))
+ module4_52_opt = self.module4_52(opt_concat_354)
+ opt_concat_360 = self.concat_360(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ module4_49_opt, module4_50_opt, module4_51_opt, module4_52_opt,
+ ))
+ module4_53_opt = self.module4_53(opt_concat_360)
+ opt_concat_366 = self.concat_366(
+ (opt_avgpool2d_269, module0_3_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ module4_43_opt, module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ module4_49_opt, module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt,
+ ))
+ module1_3_opt = self.module1_3(opt_concat_366)
+ opt_avgpool2d_369 = self.avgpool2d_369(module1_3_opt)
+ opt_flatten_370 = self.flatten_370(opt_avgpool2d_369)
+ opt_dense_371 = self.dense_371(opt_flatten_370)
+ return opt_dense_371
diff --git a/research/cvtmodel/densenet/src/densenet161.py b/research/cvtmodel/densenet/src/densenet161.py
new file mode 100644
index 0000000000000000000000000000000000000000..26977d6f0155f8efd1973b03e05fbd14a9f399d5
--- /dev/null
+++ b/research/cvtmodel/densenet/src/densenet161.py
@@ -0,0 +1,741 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module0(nn.Cell):
+ def __init__(self, batchnorm2d_1_num_features, conv2d_3_in_channels):
+ super(Module0, self).__init__()
+ self.concat_0 = P.Concat(axis=1)
+ self.batchnorm2d_1 = nn.BatchNorm2d(num_features=batchnorm2d_1_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=192,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_4 = nn.ReLU()
+ self.conv2d_5 = nn.Conv2d(in_channels=192,
+ out_channels=48,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ opt_concat_0 = self.concat_0((x,))
+ opt_batchnorm2d_1 = self.batchnorm2d_1(opt_concat_0)
+ opt_relu_2 = self.relu_2(opt_batchnorm2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_relu_4 = self.relu_4(opt_conv2d_3)
+ opt_conv2d_5 = self.conv2d_5(opt_relu_4)
+ return opt_conv2d_5
+
+
+class Module1(nn.Cell):
+ def __init__(self, batchnorm2d_0_num_features):
+ super(Module1, self).__init__()
+ self.batchnorm2d_0 = nn.BatchNorm2d(num_features=batchnorm2d_0_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_batchnorm2d_0 = self.batchnorm2d_0(x)
+ opt_relu_1 = self.relu_1(opt_batchnorm2d_0)
+ return opt_relu_1
+
+
+class Module4(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, module1_0_batchnorm2d_0_num_features):
+ super(Module4, self).__init__()
+ self.module1_0 = Module1(batchnorm2d_0_num_features=module1_0_batchnorm2d_0_num_features)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=192,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.conv2d_2 = nn.Conv2d(in_channels=192,
+ out_channels=48,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ module1_0_opt = self.module1_0(x)
+ opt_conv2d_0 = self.conv2d_0(module1_0_opt)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_conv2d_2 = self.conv2d_2(opt_relu_1)
+ return opt_conv2d_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=96,
+ kernel_size=(7, 7),
+ stride=(2, 2),
+ padding=(3, 3, 3, 3),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.pad_maxpool2d_2 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_2 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module0_0 = Module0(batchnorm2d_1_num_features=96, conv2d_3_in_channels=96)
+ self.concat_9 = P.Concat(axis=1)
+ self.module4_0 = Module4(conv2d_0_in_channels=144, module1_0_batchnorm2d_0_num_features=144)
+ self.concat_15 = P.Concat(axis=1)
+ self.module4_1 = Module4(conv2d_0_in_channels=192, module1_0_batchnorm2d_0_num_features=192)
+ self.concat_21 = P.Concat(axis=1)
+ self.module4_2 = Module4(conv2d_0_in_channels=240, module1_0_batchnorm2d_0_num_features=240)
+ self.concat_27 = P.Concat(axis=1)
+ self.module4_3 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_33 = P.Concat(axis=1)
+ self.module4_4 = Module4(conv2d_0_in_channels=336, module1_0_batchnorm2d_0_num_features=336)
+ self.concat_39 = P.Concat(axis=1)
+ self.module1_0 = Module1(batchnorm2d_0_num_features=384)
+ self.conv2d_42 = nn.Conv2d(in_channels=384,
+ out_channels=192,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_43 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_43 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_1 = Module0(batchnorm2d_1_num_features=192, conv2d_3_in_channels=192)
+ self.concat_50 = P.Concat(axis=1)
+ self.module4_5 = Module4(conv2d_0_in_channels=240, module1_0_batchnorm2d_0_num_features=240)
+ self.concat_56 = P.Concat(axis=1)
+ self.module4_6 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_62 = P.Concat(axis=1)
+ self.module4_7 = Module4(conv2d_0_in_channels=336, module1_0_batchnorm2d_0_num_features=336)
+ self.concat_68 = P.Concat(axis=1)
+ self.module4_8 = Module4(conv2d_0_in_channels=384, module1_0_batchnorm2d_0_num_features=384)
+ self.concat_74 = P.Concat(axis=1)
+ self.module4_9 = Module4(conv2d_0_in_channels=432, module1_0_batchnorm2d_0_num_features=432)
+ self.concat_80 = P.Concat(axis=1)
+ self.module4_10 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_86 = P.Concat(axis=1)
+ self.module4_11 = Module4(conv2d_0_in_channels=528, module1_0_batchnorm2d_0_num_features=528)
+ self.concat_92 = P.Concat(axis=1)
+ self.module4_12 = Module4(conv2d_0_in_channels=576, module1_0_batchnorm2d_0_num_features=576)
+ self.concat_98 = P.Concat(axis=1)
+ self.module4_13 = Module4(conv2d_0_in_channels=624, module1_0_batchnorm2d_0_num_features=624)
+ self.concat_104 = P.Concat(axis=1)
+ self.module4_14 = Module4(conv2d_0_in_channels=672, module1_0_batchnorm2d_0_num_features=672)
+ self.concat_110 = P.Concat(axis=1)
+ self.module4_15 = Module4(conv2d_0_in_channels=720, module1_0_batchnorm2d_0_num_features=720)
+ self.concat_116 = P.Concat(axis=1)
+ self.module1_1 = Module1(batchnorm2d_0_num_features=768)
+ self.conv2d_119 = nn.Conv2d(in_channels=768,
+ out_channels=384,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_120 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_120 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_2 = Module0(batchnorm2d_1_num_features=384, conv2d_3_in_channels=384)
+ self.concat_127 = P.Concat(axis=1)
+ self.module4_16 = Module4(conv2d_0_in_channels=432, module1_0_batchnorm2d_0_num_features=432)
+ self.concat_133 = P.Concat(axis=1)
+ self.module4_17 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_139 = P.Concat(axis=1)
+ self.module4_18 = Module4(conv2d_0_in_channels=528, module1_0_batchnorm2d_0_num_features=528)
+ self.concat_145 = P.Concat(axis=1)
+ self.module4_19 = Module4(conv2d_0_in_channels=576, module1_0_batchnorm2d_0_num_features=576)
+ self.concat_151 = P.Concat(axis=1)
+ self.module4_20 = Module4(conv2d_0_in_channels=624, module1_0_batchnorm2d_0_num_features=624)
+ self.concat_157 = P.Concat(axis=1)
+ self.module4_21 = Module4(conv2d_0_in_channels=672, module1_0_batchnorm2d_0_num_features=672)
+ self.concat_163 = P.Concat(axis=1)
+ self.module4_22 = Module4(conv2d_0_in_channels=720, module1_0_batchnorm2d_0_num_features=720)
+ self.concat_169 = P.Concat(axis=1)
+ self.module4_23 = Module4(conv2d_0_in_channels=768, module1_0_batchnorm2d_0_num_features=768)
+ self.concat_175 = P.Concat(axis=1)
+ self.module4_24 = Module4(conv2d_0_in_channels=816, module1_0_batchnorm2d_0_num_features=816)
+ self.concat_181 = P.Concat(axis=1)
+ self.module4_25 = Module4(conv2d_0_in_channels=864, module1_0_batchnorm2d_0_num_features=864)
+ self.concat_187 = P.Concat(axis=1)
+ self.module4_26 = Module4(conv2d_0_in_channels=912, module1_0_batchnorm2d_0_num_features=912)
+ self.concat_193 = P.Concat(axis=1)
+ self.module4_27 = Module4(conv2d_0_in_channels=960, module1_0_batchnorm2d_0_num_features=960)
+ self.concat_199 = P.Concat(axis=1)
+ self.module4_28 = Module4(conv2d_0_in_channels=1008, module1_0_batchnorm2d_0_num_features=1008)
+ self.concat_205 = P.Concat(axis=1)
+ self.module4_29 = Module4(conv2d_0_in_channels=1056, module1_0_batchnorm2d_0_num_features=1056)
+ self.concat_211 = P.Concat(axis=1)
+ self.module4_30 = Module4(conv2d_0_in_channels=1104, module1_0_batchnorm2d_0_num_features=1104)
+ self.concat_217 = P.Concat(axis=1)
+ self.module4_31 = Module4(conv2d_0_in_channels=1152, module1_0_batchnorm2d_0_num_features=1152)
+ self.concat_223 = P.Concat(axis=1)
+ self.module4_32 = Module4(conv2d_0_in_channels=1200, module1_0_batchnorm2d_0_num_features=1200)
+ self.concat_229 = P.Concat(axis=1)
+ self.module4_33 = Module4(conv2d_0_in_channels=1248, module1_0_batchnorm2d_0_num_features=1248)
+ self.concat_235 = P.Concat(axis=1)
+ self.module4_34 = Module4(conv2d_0_in_channels=1296, module1_0_batchnorm2d_0_num_features=1296)
+ self.concat_241 = P.Concat(axis=1)
+ self.module4_35 = Module4(conv2d_0_in_channels=1344, module1_0_batchnorm2d_0_num_features=1344)
+ self.concat_247 = P.Concat(axis=1)
+ self.module4_36 = Module4(conv2d_0_in_channels=1392, module1_0_batchnorm2d_0_num_features=1392)
+ self.concat_253 = P.Concat(axis=1)
+ self.module4_37 = Module4(conv2d_0_in_channels=1440, module1_0_batchnorm2d_0_num_features=1440)
+ self.concat_259 = P.Concat(axis=1)
+ self.module4_38 = Module4(conv2d_0_in_channels=1488, module1_0_batchnorm2d_0_num_features=1488)
+ self.concat_265 = P.Concat(axis=1)
+ self.module4_39 = Module4(conv2d_0_in_channels=1536, module1_0_batchnorm2d_0_num_features=1536)
+ self.concat_271 = P.Concat(axis=1)
+ self.module4_40 = Module4(conv2d_0_in_channels=1584, module1_0_batchnorm2d_0_num_features=1584)
+ self.concat_277 = P.Concat(axis=1)
+ self.module4_41 = Module4(conv2d_0_in_channels=1632, module1_0_batchnorm2d_0_num_features=1632)
+ self.concat_283 = P.Concat(axis=1)
+ self.module4_42 = Module4(conv2d_0_in_channels=1680, module1_0_batchnorm2d_0_num_features=1680)
+ self.concat_289 = P.Concat(axis=1)
+ self.module4_43 = Module4(conv2d_0_in_channels=1728, module1_0_batchnorm2d_0_num_features=1728)
+ self.concat_295 = P.Concat(axis=1)
+ self.module4_44 = Module4(conv2d_0_in_channels=1776, module1_0_batchnorm2d_0_num_features=1776)
+ self.concat_301 = P.Concat(axis=1)
+ self.module4_45 = Module4(conv2d_0_in_channels=1824, module1_0_batchnorm2d_0_num_features=1824)
+ self.concat_307 = P.Concat(axis=1)
+ self.module4_46 = Module4(conv2d_0_in_channels=1872, module1_0_batchnorm2d_0_num_features=1872)
+ self.concat_313 = P.Concat(axis=1)
+ self.module4_47 = Module4(conv2d_0_in_channels=1920, module1_0_batchnorm2d_0_num_features=1920)
+ self.concat_319 = P.Concat(axis=1)
+ self.module4_48 = Module4(conv2d_0_in_channels=1968, module1_0_batchnorm2d_0_num_features=1968)
+ self.concat_325 = P.Concat(axis=1)
+ self.module4_49 = Module4(conv2d_0_in_channels=2016, module1_0_batchnorm2d_0_num_features=2016)
+ self.concat_331 = P.Concat(axis=1)
+ self.module4_50 = Module4(conv2d_0_in_channels=2064, module1_0_batchnorm2d_0_num_features=2064)
+ self.concat_337 = P.Concat(axis=1)
+ self.module1_2 = Module1(batchnorm2d_0_num_features=2112)
+ self.conv2d_340 = nn.Conv2d(in_channels=2112,
+ out_channels=1056,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_341 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_341 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_3 = Module0(batchnorm2d_1_num_features=1056, conv2d_3_in_channels=1056)
+ self.concat_348 = P.Concat(axis=1)
+ self.module4_51 = Module4(conv2d_0_in_channels=1104, module1_0_batchnorm2d_0_num_features=1104)
+ self.concat_354 = P.Concat(axis=1)
+ self.module4_52 = Module4(conv2d_0_in_channels=1152, module1_0_batchnorm2d_0_num_features=1152)
+ self.concat_360 = P.Concat(axis=1)
+ self.module4_53 = Module4(conv2d_0_in_channels=1200, module1_0_batchnorm2d_0_num_features=1200)
+ self.concat_366 = P.Concat(axis=1)
+ self.module4_54 = Module4(conv2d_0_in_channels=1248, module1_0_batchnorm2d_0_num_features=1248)
+ self.concat_372 = P.Concat(axis=1)
+ self.module4_55 = Module4(conv2d_0_in_channels=1296, module1_0_batchnorm2d_0_num_features=1296)
+ self.concat_378 = P.Concat(axis=1)
+ self.module4_56 = Module4(conv2d_0_in_channels=1344, module1_0_batchnorm2d_0_num_features=1344)
+ self.concat_384 = P.Concat(axis=1)
+ self.module4_57 = Module4(conv2d_0_in_channels=1392, module1_0_batchnorm2d_0_num_features=1392)
+ self.concat_390 = P.Concat(axis=1)
+ self.module4_58 = Module4(conv2d_0_in_channels=1440, module1_0_batchnorm2d_0_num_features=1440)
+ self.concat_396 = P.Concat(axis=1)
+ self.module4_59 = Module4(conv2d_0_in_channels=1488, module1_0_batchnorm2d_0_num_features=1488)
+ self.concat_402 = P.Concat(axis=1)
+ self.module4_60 = Module4(conv2d_0_in_channels=1536, module1_0_batchnorm2d_0_num_features=1536)
+ self.concat_408 = P.Concat(axis=1)
+ self.module4_61 = Module4(conv2d_0_in_channels=1584, module1_0_batchnorm2d_0_num_features=1584)
+ self.concat_414 = P.Concat(axis=1)
+ self.module4_62 = Module4(conv2d_0_in_channels=1632, module1_0_batchnorm2d_0_num_features=1632)
+ self.concat_420 = P.Concat(axis=1)
+ self.module4_63 = Module4(conv2d_0_in_channels=1680, module1_0_batchnorm2d_0_num_features=1680)
+ self.concat_426 = P.Concat(axis=1)
+ self.module4_64 = Module4(conv2d_0_in_channels=1728, module1_0_batchnorm2d_0_num_features=1728)
+ self.concat_432 = P.Concat(axis=1)
+ self.module4_65 = Module4(conv2d_0_in_channels=1776, module1_0_batchnorm2d_0_num_features=1776)
+ self.concat_438 = P.Concat(axis=1)
+ self.module4_66 = Module4(conv2d_0_in_channels=1824, module1_0_batchnorm2d_0_num_features=1824)
+ self.concat_444 = P.Concat(axis=1)
+ self.module4_67 = Module4(conv2d_0_in_channels=1872, module1_0_batchnorm2d_0_num_features=1872)
+ self.concat_450 = P.Concat(axis=1)
+ self.module4_68 = Module4(conv2d_0_in_channels=1920, module1_0_batchnorm2d_0_num_features=1920)
+ self.concat_456 = P.Concat(axis=1)
+ self.module4_69 = Module4(conv2d_0_in_channels=1968, module1_0_batchnorm2d_0_num_features=1968)
+ self.concat_462 = P.Concat(axis=1)
+ self.module4_70 = Module4(conv2d_0_in_channels=2016, module1_0_batchnorm2d_0_num_features=2016)
+ self.concat_468 = P.Concat(axis=1)
+ self.module4_71 = Module4(conv2d_0_in_channels=2064, module1_0_batchnorm2d_0_num_features=2064)
+ self.concat_474 = P.Concat(axis=1)
+ self.module4_72 = Module4(conv2d_0_in_channels=2112, module1_0_batchnorm2d_0_num_features=2112)
+ self.concat_480 = P.Concat(axis=1)
+ self.module4_73 = Module4(conv2d_0_in_channels=2160, module1_0_batchnorm2d_0_num_features=2160)
+ self.concat_486 = P.Concat(axis=1)
+ self.module1_3 = Module1(batchnorm2d_0_num_features=2208)
+ self.avgpool2d_489 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.flatten_490 = nn.Flatten()
+ self.dense_491 = nn.Dense(in_channels=2208, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_maxpool2d_2 = self.pad_maxpool2d_2(opt_relu_1)
+ opt_maxpool2d_2 = self.maxpool2d_2(opt_maxpool2d_2)
+ module0_0_opt = self.module0_0(opt_maxpool2d_2)
+ opt_concat_9 = self.concat_9((opt_maxpool2d_2, module0_0_opt,))
+ module4_0_opt = self.module4_0(opt_concat_9)
+ opt_concat_15 = self.concat_15((opt_maxpool2d_2, module0_0_opt, module4_0_opt,))
+ module4_1_opt = self.module4_1(opt_concat_15)
+ opt_concat_21 = self.concat_21((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt,))
+ module4_2_opt = self.module4_2(opt_concat_21)
+ opt_concat_27 = self.concat_27((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt,))
+ module4_3_opt = self.module4_3(opt_concat_27)
+ opt_concat_33 = self.concat_33(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt,
+ ))
+ module4_4_opt = self.module4_4(opt_concat_33)
+ opt_concat_39 = self.concat_39(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt, module4_4_opt,
+ ))
+ module1_0_opt = self.module1_0(opt_concat_39)
+ opt_conv2d_42 = self.conv2d_42(module1_0_opt)
+ opt_avgpool2d_43 = self.pad_avgpool2d_43(opt_conv2d_42)
+ opt_avgpool2d_43 = self.avgpool2d_43(opt_avgpool2d_43)
+ module0_1_opt = self.module0_1(opt_avgpool2d_43)
+ opt_concat_50 = self.concat_50((opt_avgpool2d_43, module0_1_opt,))
+ module4_5_opt = self.module4_5(opt_concat_50)
+ opt_concat_56 = self.concat_56((opt_avgpool2d_43, module0_1_opt, module4_5_opt,))
+ module4_6_opt = self.module4_6(opt_concat_56)
+ opt_concat_62 = self.concat_62((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt,))
+ module4_7_opt = self.module4_7(opt_concat_62)
+ opt_concat_68 = self.concat_68((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,))
+ module4_8_opt = self.module4_8(opt_concat_68)
+ opt_concat_74 = self.concat_74(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt,
+ ))
+ module4_9_opt = self.module4_9(opt_concat_74)
+ opt_concat_80 = self.concat_80(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ ))
+ module4_10_opt = self.module4_10(opt_concat_80)
+ opt_concat_86 = self.concat_86((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt,
+ ))
+ module4_11_opt = self.module4_11(opt_concat_86)
+ opt_concat_92 = self.concat_92((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt,
+ ))
+ module4_12_opt = self.module4_12(opt_concat_92)
+ opt_concat_98 = self.concat_98((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt, module4_12_opt,
+ ))
+ module4_13_opt = self.module4_13(opt_concat_98)
+ opt_concat_104 = self.concat_104(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt,
+ ))
+ module4_14_opt = self.module4_14(opt_concat_104)
+ opt_concat_110 = self.concat_110(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt,
+ ))
+ module4_15_opt = self.module4_15(opt_concat_110)
+ opt_concat_116 = self.concat_116(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt, module4_15_opt,
+ ))
+ module1_1_opt = self.module1_1(opt_concat_116)
+ opt_conv2d_119 = self.conv2d_119(module1_1_opt)
+ opt_avgpool2d_120 = self.pad_avgpool2d_120(opt_conv2d_119)
+ opt_avgpool2d_120 = self.avgpool2d_120(opt_avgpool2d_120)
+ module0_2_opt = self.module0_2(opt_avgpool2d_120)
+ opt_concat_127 = self.concat_127((opt_avgpool2d_120, module0_2_opt,))
+ module4_16_opt = self.module4_16(opt_concat_127)
+ opt_concat_133 = self.concat_133((opt_avgpool2d_120, module0_2_opt, module4_16_opt,))
+ module4_17_opt = self.module4_17(opt_concat_133)
+ opt_concat_139 = self.concat_139((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,))
+ module4_18_opt = self.module4_18(opt_concat_139)
+ opt_concat_145 = self.concat_145(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt,
+ ))
+ module4_19_opt = self.module4_19(opt_concat_145)
+ opt_concat_151 = self.concat_151(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ ))
+ module4_20_opt = self.module4_20(opt_concat_151)
+ opt_concat_157 = self.concat_157((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt,
+ ))
+ module4_21_opt = self.module4_21(opt_concat_157)
+ opt_concat_163 = self.concat_163((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt, module4_21_opt,
+ ))
+ module4_22_opt = self.module4_22(opt_concat_163)
+ opt_concat_169 = self.concat_169(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt,
+ ))
+ module4_23_opt = self.module4_23(opt_concat_169)
+ opt_concat_175 = self.concat_175(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt,
+ ))
+ module4_24_opt = self.module4_24(opt_concat_175)
+ opt_concat_181 = self.concat_181(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt,
+ ))
+ module4_25_opt = self.module4_25(opt_concat_181)
+ opt_concat_187 = self.concat_187(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ ))
+ module4_26_opt = self.module4_26(opt_concat_187)
+ opt_concat_193 = self.concat_193(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt,
+ ))
+ module4_27_opt = self.module4_27(opt_concat_193)
+ opt_concat_199 = self.concat_199(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt,
+ ))
+ module4_28_opt = self.module4_28(opt_concat_199)
+ opt_concat_205 = self.concat_205(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt,
+ ))
+ module4_29_opt = self.module4_29(opt_concat_205)
+ opt_concat_211 = self.concat_211(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt,
+ ))
+ module4_30_opt = self.module4_30(opt_concat_211)
+ opt_concat_217 = self.concat_217(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt,
+ ))
+ module4_31_opt = self.module4_31(opt_concat_217)
+ opt_concat_223 = self.concat_223(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ ))
+ module4_32_opt = self.module4_32(opt_concat_223)
+ opt_concat_229 = self.concat_229(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt,
+ ))
+ module4_33_opt = self.module4_33(opt_concat_229)
+ opt_concat_235 = self.concat_235(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt,
+ ))
+ module4_34_opt = self.module4_34(opt_concat_235)
+ opt_concat_241 = self.concat_241(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt,
+ ))
+ module4_35_opt = self.module4_35(opt_concat_241)
+ opt_concat_247 = self.concat_247(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt,
+ ))
+ module4_36_opt = self.module4_36(opt_concat_247)
+ opt_concat_253 = self.concat_253(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt,
+ ))
+ module4_37_opt = self.module4_37(opt_concat_253)
+ opt_concat_259 = self.concat_259(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ ))
+ module4_38_opt = self.module4_38(opt_concat_259)
+ opt_concat_265 = self.concat_265(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt,
+ ))
+ module4_39_opt = self.module4_39(opt_concat_265)
+ opt_concat_271 = self.concat_271(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt,
+ ))
+ module4_40_opt = self.module4_40(opt_concat_271)
+ opt_concat_277 = self.concat_277(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt,
+ ))
+ module4_41_opt = self.module4_41(opt_concat_277)
+ opt_concat_283 = self.concat_283(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt,
+ ))
+ module4_42_opt = self.module4_42(opt_concat_283)
+ opt_concat_289 = self.concat_289(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ ))
+ module4_43_opt = self.module4_43(opt_concat_289)
+ opt_concat_295 = self.concat_295(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ ))
+ module4_44_opt = self.module4_44(opt_concat_295)
+ opt_concat_301 = self.concat_301(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt,
+ ))
+ module4_45_opt = self.module4_45(opt_concat_301)
+ opt_concat_307 = self.concat_307(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt,
+ ))
+ module4_46_opt = self.module4_46(opt_concat_307)
+ opt_concat_313 = self.concat_313(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt,
+ ))
+ module4_47_opt = self.module4_47(opt_concat_313)
+ opt_concat_319 = self.concat_319(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt,
+ ))
+ module4_48_opt = self.module4_48(opt_concat_319)
+ opt_concat_325 = self.concat_325(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ ))
+ module4_49_opt = self.module4_49(opt_concat_325)
+ opt_concat_331 = self.concat_331(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ ))
+ module4_50_opt = self.module4_50(opt_concat_331)
+ opt_concat_337 = self.concat_337(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt,
+ ))
+ module1_2_opt = self.module1_2(opt_concat_337)
+ opt_conv2d_340 = self.conv2d_340(module1_2_opt)
+ opt_avgpool2d_341 = self.pad_avgpool2d_341(opt_conv2d_340)
+ opt_avgpool2d_341 = self.avgpool2d_341(opt_avgpool2d_341)
+ module0_3_opt = self.module0_3(opt_avgpool2d_341)
+ opt_concat_348 = self.concat_348((opt_avgpool2d_341, module0_3_opt,))
+ module4_51_opt = self.module4_51(opt_concat_348)
+ opt_concat_354 = self.concat_354((opt_avgpool2d_341, module0_3_opt, module4_51_opt,))
+ module4_52_opt = self.module4_52(opt_concat_354)
+ opt_concat_360 = self.concat_360((opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt,))
+ module4_53_opt = self.module4_53(opt_concat_360)
+ opt_concat_366 = self.concat_366(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt,
+ ))
+ module4_54_opt = self.module4_54(opt_concat_366)
+ opt_concat_372 = self.concat_372(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ ))
+ module4_55_opt = self.module4_55(opt_concat_372)
+ opt_concat_378 = self.concat_378((opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt,
+ module4_53_opt, module4_54_opt, module4_55_opt,
+ ))
+ module4_56_opt = self.module4_56(opt_concat_378)
+ opt_concat_384 = self.concat_384((opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt,
+ module4_53_opt, module4_54_opt, module4_55_opt, module4_56_opt,
+ ))
+ module4_57_opt = self.module4_57(opt_concat_384)
+ opt_concat_390 = self.concat_390(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt,
+ ))
+ module4_58_opt = self.module4_58(opt_concat_390)
+ opt_concat_396 = self.concat_396(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt,
+ ))
+ module4_59_opt = self.module4_59(opt_concat_396)
+ opt_concat_402 = self.concat_402(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt,
+ ))
+ module4_60_opt = self.module4_60(opt_concat_402)
+ opt_concat_408 = self.concat_408(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ ))
+ module4_61_opt = self.module4_61(opt_concat_408)
+ opt_concat_414 = self.concat_414(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt,
+ ))
+ module4_62_opt = self.module4_62(opt_concat_414)
+ opt_concat_420 = self.concat_420(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt,
+ ))
+ module4_63_opt = self.module4_63(opt_concat_420)
+ opt_concat_426 = self.concat_426(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt,
+ ))
+ module4_64_opt = self.module4_64(opt_concat_426)
+ opt_concat_432 = self.concat_432(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt,
+ ))
+ module4_65_opt = self.module4_65(opt_concat_432)
+ opt_concat_438 = self.concat_438(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt,
+ ))
+ module4_66_opt = self.module4_66(opt_concat_438)
+ opt_concat_444 = self.concat_444(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ ))
+ module4_67_opt = self.module4_67(opt_concat_444)
+ opt_concat_450 = self.concat_450(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt,
+ ))
+ module4_68_opt = self.module4_68(opt_concat_450)
+ opt_concat_456 = self.concat_456(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt,
+ ))
+ module4_69_opt = self.module4_69(opt_concat_456)
+ opt_concat_462 = self.concat_462(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt,
+ ))
+ module4_70_opt = self.module4_70(opt_concat_462)
+ opt_concat_468 = self.concat_468(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt,
+ ))
+ module4_71_opt = self.module4_71(opt_concat_468)
+ opt_concat_474 = self.concat_474(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt,
+ ))
+ module4_72_opt = self.module4_72(opt_concat_474)
+ opt_concat_480 = self.concat_480(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ ))
+ module4_73_opt = self.module4_73(opt_concat_480)
+ opt_concat_486 = self.concat_486(
+ (opt_avgpool2d_341, module0_3_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ module4_55_opt, module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ module4_61_opt, module4_62_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt,
+ ))
+ module1_3_opt = self.module1_3(opt_concat_486)
+ opt_avgpool2d_489 = self.avgpool2d_489(module1_3_opt)
+ opt_flatten_490 = self.flatten_490(opt_avgpool2d_489)
+ opt_dense_491 = self.dense_491(opt_flatten_490)
+ return opt_dense_491
diff --git a/research/cvtmodel/densenet/src/densenet169.py b/research/cvtmodel/densenet/src/densenet169.py
new file mode 100644
index 0000000000000000000000000000000000000000..33b2f2fa80888ce4b2c872f31d6f6f618551544b
--- /dev/null
+++ b/research/cvtmodel/densenet/src/densenet169.py
@@ -0,0 +1,1462 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module0(nn.Cell):
+ def __init__(self, batchnorm2d_1_num_features, conv2d_3_in_channels):
+ super(Module0, self).__init__()
+ self.concat_0 = P.Concat(axis=1)
+ self.batchnorm2d_1 = nn.BatchNorm2d(num_features=batchnorm2d_1_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_4 = nn.ReLU()
+ self.conv2d_5 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ opt_concat_0 = self.concat_0((x,))
+ opt_batchnorm2d_1 = self.batchnorm2d_1(opt_concat_0)
+ opt_relu_2 = self.relu_2(opt_batchnorm2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_relu_4 = self.relu_4(opt_conv2d_3)
+ opt_conv2d_5 = self.conv2d_5(opt_relu_4)
+ return opt_conv2d_5
+
+
+class Module21(nn.Cell):
+ def __init__(self, batchnorm2d_0_num_features):
+ super(Module21, self).__init__()
+ self.batchnorm2d_0 = nn.BatchNorm2d(num_features=batchnorm2d_0_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_batchnorm2d_0 = self.batchnorm2d_0(x)
+ opt_relu_1 = self.relu_1(opt_batchnorm2d_0)
+ return opt_relu_1
+
+
+class Module24(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, module21_0_batchnorm2d_0_num_features):
+ super(Module24, self).__init__()
+ self.module21_0 = Module21(batchnorm2d_0_num_features=module21_0_batchnorm2d_0_num_features)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.conv2d_2 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ module21_0_opt = self.module21_0(x)
+ opt_conv2d_0 = self.conv2d_0(module21_0_opt)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_conv2d_2 = self.conv2d_2(opt_relu_1)
+ return opt_conv2d_2
+
+
+class Module26(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, module21_0_batchnorm2d_0_num_features):
+ super(Module26, self).__init__()
+ self.module21_0 = Module21(batchnorm2d_0_num_features=module21_0_batchnorm2d_0_num_features)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ module21_0_opt = self.module21_0(x)
+ opt_conv2d_0 = self.conv2d_0(module21_0_opt)
+ return opt_conv2d_0
+
+
+class Module20(nn.Cell):
+ def __init__(
+ self, batchnorm2d_2_num_features, conv2d_4_in_channels, batchnorm2d_8_num_features, conv2d_10_in_channels,
+ batchnorm2d_14_num_features, conv2d_16_in_channels, batchnorm2d_20_num_features, conv2d_22_in_channels,
+ batchnorm2d_26_num_features, conv2d_28_in_channels, batchnorm2d_32_num_features, conv2d_34_in_channels,
+ batchnorm2d_38_num_features, conv2d_40_in_channels, batchnorm2d_44_num_features, conv2d_46_in_channels,
+ batchnorm2d_50_num_features, conv2d_52_in_channels, batchnorm2d_56_num_features, conv2d_58_in_channels,
+ batchnorm2d_62_num_features, conv2d_64_in_channels, batchnorm2d_68_num_features, conv2d_70_in_channels,
+ batchnorm2d_74_num_features, conv2d_76_in_channels, batchnorm2d_80_num_features, conv2d_82_in_channels,
+ batchnorm2d_86_num_features, conv2d_88_in_channels, batchnorm2d_92_num_features, conv2d_94_in_channels,
+ batchnorm2d_98_num_features, conv2d_100_in_channels, batchnorm2d_104_num_features, conv2d_106_in_channels,
+ batchnorm2d_110_num_features, conv2d_112_in_channels, batchnorm2d_116_num_features, conv2d_118_in_channels,
+ batchnorm2d_122_num_features, conv2d_124_in_channels, batchnorm2d_128_num_features, conv2d_130_in_channels,
+ batchnorm2d_134_num_features, conv2d_136_in_channels, batchnorm2d_140_num_features, conv2d_142_in_channels,
+ batchnorm2d_146_num_features, conv2d_148_in_channels, batchnorm2d_152_num_features, conv2d_154_in_channels,
+ batchnorm2d_158_num_features, conv2d_160_in_channels, batchnorm2d_164_num_features, conv2d_166_in_channels,
+ batchnorm2d_170_num_features, conv2d_172_in_channels, batchnorm2d_176_num_features, conv2d_178_in_channels,
+ batchnorm2d_182_num_features, conv2d_184_in_channels, module0_0_batchnorm2d_1_num_features,
+ module0_0_conv2d_3_in_channels):
+ super(Module20, self).__init__()
+ self.pad_avgpool2d_0 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_0 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_0 = Module0(batchnorm2d_1_num_features=module0_0_batchnorm2d_1_num_features,
+ conv2d_3_in_channels=module0_0_conv2d_3_in_channels)
+ self.concat_1 = P.Concat(axis=1)
+ self.batchnorm2d_2 = nn.BatchNorm2d(num_features=batchnorm2d_2_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_3 = nn.ReLU()
+ self.conv2d_4 = nn.Conv2d(in_channels=conv2d_4_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_5 = nn.ReLU()
+ self.conv2d_6 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_7 = P.Concat(axis=1)
+ self.batchnorm2d_8 = nn.BatchNorm2d(num_features=batchnorm2d_8_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_9 = nn.ReLU()
+ self.conv2d_10 = nn.Conv2d(in_channels=conv2d_10_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_11 = nn.ReLU()
+ self.conv2d_12 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_13 = P.Concat(axis=1)
+ self.batchnorm2d_14 = nn.BatchNorm2d(num_features=batchnorm2d_14_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_15 = nn.ReLU()
+ self.conv2d_16 = nn.Conv2d(in_channels=conv2d_16_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_17 = nn.ReLU()
+ self.conv2d_18 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_19 = P.Concat(axis=1)
+ self.batchnorm2d_20 = nn.BatchNorm2d(num_features=batchnorm2d_20_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_21 = nn.ReLU()
+ self.conv2d_22 = nn.Conv2d(in_channels=conv2d_22_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_23 = nn.ReLU()
+ self.conv2d_24 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_25 = P.Concat(axis=1)
+ self.batchnorm2d_26 = nn.BatchNorm2d(num_features=batchnorm2d_26_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_27 = nn.ReLU()
+ self.conv2d_28 = nn.Conv2d(in_channels=conv2d_28_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_29 = nn.ReLU()
+ self.conv2d_30 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_31 = P.Concat(axis=1)
+ self.batchnorm2d_32 = nn.BatchNorm2d(num_features=batchnorm2d_32_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_33 = nn.ReLU()
+ self.conv2d_34 = nn.Conv2d(in_channels=conv2d_34_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_35 = nn.ReLU()
+ self.conv2d_36 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_37 = P.Concat(axis=1)
+ self.batchnorm2d_38 = nn.BatchNorm2d(num_features=batchnorm2d_38_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_39 = nn.ReLU()
+ self.conv2d_40 = nn.Conv2d(in_channels=conv2d_40_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_41 = nn.ReLU()
+ self.conv2d_42 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_43 = P.Concat(axis=1)
+ self.batchnorm2d_44 = nn.BatchNorm2d(num_features=batchnorm2d_44_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_45 = nn.ReLU()
+ self.conv2d_46 = nn.Conv2d(in_channels=conv2d_46_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_47 = nn.ReLU()
+ self.conv2d_48 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_49 = P.Concat(axis=1)
+ self.batchnorm2d_50 = nn.BatchNorm2d(num_features=batchnorm2d_50_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_51 = nn.ReLU()
+ self.conv2d_52 = nn.Conv2d(in_channels=conv2d_52_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_53 = nn.ReLU()
+ self.conv2d_54 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_55 = P.Concat(axis=1)
+ self.batchnorm2d_56 = nn.BatchNorm2d(num_features=batchnorm2d_56_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_57 = nn.ReLU()
+ self.conv2d_58 = nn.Conv2d(in_channels=conv2d_58_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_59 = nn.ReLU()
+ self.conv2d_60 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_61 = P.Concat(axis=1)
+ self.batchnorm2d_62 = nn.BatchNorm2d(num_features=batchnorm2d_62_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_63 = nn.ReLU()
+ self.conv2d_64 = nn.Conv2d(in_channels=conv2d_64_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_65 = nn.ReLU()
+ self.conv2d_66 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_67 = P.Concat(axis=1)
+ self.batchnorm2d_68 = nn.BatchNorm2d(num_features=batchnorm2d_68_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_69 = nn.ReLU()
+ self.conv2d_70 = nn.Conv2d(in_channels=conv2d_70_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_71 = nn.ReLU()
+ self.conv2d_72 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_73 = P.Concat(axis=1)
+ self.batchnorm2d_74 = nn.BatchNorm2d(num_features=batchnorm2d_74_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_75 = nn.ReLU()
+ self.conv2d_76 = nn.Conv2d(in_channels=conv2d_76_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_77 = nn.ReLU()
+ self.conv2d_78 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_79 = P.Concat(axis=1)
+ self.batchnorm2d_80 = nn.BatchNorm2d(num_features=batchnorm2d_80_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_81 = nn.ReLU()
+ self.conv2d_82 = nn.Conv2d(in_channels=conv2d_82_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_83 = nn.ReLU()
+ self.conv2d_84 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_85 = P.Concat(axis=1)
+ self.batchnorm2d_86 = nn.BatchNorm2d(num_features=batchnorm2d_86_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_87 = nn.ReLU()
+ self.conv2d_88 = nn.Conv2d(in_channels=conv2d_88_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_89 = nn.ReLU()
+ self.conv2d_90 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_91 = P.Concat(axis=1)
+ self.batchnorm2d_92 = nn.BatchNorm2d(num_features=batchnorm2d_92_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_93 = nn.ReLU()
+ self.conv2d_94 = nn.Conv2d(in_channels=conv2d_94_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_95 = nn.ReLU()
+ self.conv2d_96 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_97 = P.Concat(axis=1)
+ self.batchnorm2d_98 = nn.BatchNorm2d(num_features=batchnorm2d_98_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_99 = nn.ReLU()
+ self.conv2d_100 = nn.Conv2d(in_channels=conv2d_100_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_101 = nn.ReLU()
+ self.conv2d_102 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_103 = P.Concat(axis=1)
+ self.batchnorm2d_104 = nn.BatchNorm2d(num_features=batchnorm2d_104_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_105 = nn.ReLU()
+ self.conv2d_106 = nn.Conv2d(in_channels=conv2d_106_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_107 = nn.ReLU()
+ self.conv2d_108 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_109 = P.Concat(axis=1)
+ self.batchnorm2d_110 = nn.BatchNorm2d(num_features=batchnorm2d_110_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_111 = nn.ReLU()
+ self.conv2d_112 = nn.Conv2d(in_channels=conv2d_112_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_113 = nn.ReLU()
+ self.conv2d_114 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_115 = P.Concat(axis=1)
+ self.batchnorm2d_116 = nn.BatchNorm2d(num_features=batchnorm2d_116_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_117 = nn.ReLU()
+ self.conv2d_118 = nn.Conv2d(in_channels=conv2d_118_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_119 = nn.ReLU()
+ self.conv2d_120 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_121 = P.Concat(axis=1)
+ self.batchnorm2d_122 = nn.BatchNorm2d(num_features=batchnorm2d_122_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_123 = nn.ReLU()
+ self.conv2d_124 = nn.Conv2d(in_channels=conv2d_124_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_125 = nn.ReLU()
+ self.conv2d_126 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_127 = P.Concat(axis=1)
+ self.batchnorm2d_128 = nn.BatchNorm2d(num_features=batchnorm2d_128_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_129 = nn.ReLU()
+ self.conv2d_130 = nn.Conv2d(in_channels=conv2d_130_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_131 = nn.ReLU()
+ self.conv2d_132 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_133 = P.Concat(axis=1)
+ self.batchnorm2d_134 = nn.BatchNorm2d(num_features=batchnorm2d_134_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_135 = nn.ReLU()
+ self.conv2d_136 = nn.Conv2d(in_channels=conv2d_136_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_137 = nn.ReLU()
+ self.conv2d_138 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_139 = P.Concat(axis=1)
+ self.batchnorm2d_140 = nn.BatchNorm2d(num_features=batchnorm2d_140_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_141 = nn.ReLU()
+ self.conv2d_142 = nn.Conv2d(in_channels=conv2d_142_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_143 = nn.ReLU()
+ self.conv2d_144 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_145 = P.Concat(axis=1)
+ self.batchnorm2d_146 = nn.BatchNorm2d(num_features=batchnorm2d_146_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_147 = nn.ReLU()
+ self.conv2d_148 = nn.Conv2d(in_channels=conv2d_148_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_149 = nn.ReLU()
+ self.conv2d_150 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_151 = P.Concat(axis=1)
+ self.batchnorm2d_152 = nn.BatchNorm2d(num_features=batchnorm2d_152_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_153 = nn.ReLU()
+ self.conv2d_154 = nn.Conv2d(in_channels=conv2d_154_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_155 = nn.ReLU()
+ self.conv2d_156 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_157 = P.Concat(axis=1)
+ self.batchnorm2d_158 = nn.BatchNorm2d(num_features=batchnorm2d_158_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_159 = nn.ReLU()
+ self.conv2d_160 = nn.Conv2d(in_channels=conv2d_160_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_161 = nn.ReLU()
+ self.conv2d_162 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_163 = P.Concat(axis=1)
+ self.batchnorm2d_164 = nn.BatchNorm2d(num_features=batchnorm2d_164_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_165 = nn.ReLU()
+ self.conv2d_166 = nn.Conv2d(in_channels=conv2d_166_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_167 = nn.ReLU()
+ self.conv2d_168 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_169 = P.Concat(axis=1)
+ self.batchnorm2d_170 = nn.BatchNorm2d(num_features=batchnorm2d_170_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_171 = nn.ReLU()
+ self.conv2d_172 = nn.Conv2d(in_channels=conv2d_172_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_173 = nn.ReLU()
+ self.conv2d_174 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_175 = P.Concat(axis=1)
+ self.batchnorm2d_176 = nn.BatchNorm2d(num_features=batchnorm2d_176_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_177 = nn.ReLU()
+ self.conv2d_178 = nn.Conv2d(in_channels=conv2d_178_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_179 = nn.ReLU()
+ self.conv2d_180 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_181 = P.Concat(axis=1)
+ self.batchnorm2d_182 = nn.BatchNorm2d(num_features=batchnorm2d_182_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_183 = nn.ReLU()
+ self.conv2d_184 = nn.Conv2d(in_channels=conv2d_184_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_185 = nn.ReLU()
+ self.conv2d_186 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.concat_187 = P.Concat(axis=1)
+
+ def construct(self, x):
+ opt_avgpool2d_0 = self.pad_avgpool2d_0(x)
+ opt_avgpool2d_0 = self.avgpool2d_0(opt_avgpool2d_0)
+ module0_0_opt = self.module0_0(opt_avgpool2d_0)
+ opt_concat_1 = self.concat_1((opt_avgpool2d_0, module0_0_opt,))
+ opt_batchnorm2d_2 = self.batchnorm2d_2(opt_concat_1)
+ opt_relu_3 = self.relu_3(opt_batchnorm2d_2)
+ opt_conv2d_4 = self.conv2d_4(opt_relu_3)
+ opt_relu_5 = self.relu_5(opt_conv2d_4)
+ opt_conv2d_6 = self.conv2d_6(opt_relu_5)
+ opt_concat_7 = self.concat_7((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6,))
+ opt_batchnorm2d_8 = self.batchnorm2d_8(opt_concat_7)
+ opt_relu_9 = self.relu_9(opt_batchnorm2d_8)
+ opt_conv2d_10 = self.conv2d_10(opt_relu_9)
+ opt_relu_11 = self.relu_11(opt_conv2d_10)
+ opt_conv2d_12 = self.conv2d_12(opt_relu_11)
+ opt_concat_13 = self.concat_13((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12,))
+ opt_batchnorm2d_14 = self.batchnorm2d_14(opt_concat_13)
+ opt_relu_15 = self.relu_15(opt_batchnorm2d_14)
+ opt_conv2d_16 = self.conv2d_16(opt_relu_15)
+ opt_relu_17 = self.relu_17(opt_conv2d_16)
+ opt_conv2d_18 = self.conv2d_18(opt_relu_17)
+ opt_concat_19 = self.concat_19((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18,))
+ opt_batchnorm2d_20 = self.batchnorm2d_20(opt_concat_19)
+ opt_relu_21 = self.relu_21(opt_batchnorm2d_20)
+ opt_conv2d_22 = self.conv2d_22(opt_relu_21)
+ opt_relu_23 = self.relu_23(opt_conv2d_22)
+ opt_conv2d_24 = self.conv2d_24(opt_relu_23)
+ opt_concat_25 = self.concat_25(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24,
+ ))
+ opt_batchnorm2d_26 = self.batchnorm2d_26(opt_concat_25)
+ opt_relu_27 = self.relu_27(opt_batchnorm2d_26)
+ opt_conv2d_28 = self.conv2d_28(opt_relu_27)
+ opt_relu_29 = self.relu_29(opt_conv2d_28)
+ opt_conv2d_30 = self.conv2d_30(opt_relu_29)
+ opt_concat_31 = self.concat_31(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ ))
+ opt_batchnorm2d_32 = self.batchnorm2d_32(opt_concat_31)
+ opt_relu_33 = self.relu_33(opt_batchnorm2d_32)
+ opt_conv2d_34 = self.conv2d_34(opt_relu_33)
+ opt_relu_35 = self.relu_35(opt_conv2d_34)
+ opt_conv2d_36 = self.conv2d_36(opt_relu_35)
+ opt_concat_37 = self.concat_37((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18,
+ opt_conv2d_24, opt_conv2d_30, opt_conv2d_36,
+ ))
+ opt_batchnorm2d_38 = self.batchnorm2d_38(opt_concat_37)
+ opt_relu_39 = self.relu_39(opt_batchnorm2d_38)
+ opt_conv2d_40 = self.conv2d_40(opt_relu_39)
+ opt_relu_41 = self.relu_41(opt_conv2d_40)
+ opt_conv2d_42 = self.conv2d_42(opt_relu_41)
+ opt_concat_43 = self.concat_43((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18,
+ opt_conv2d_24, opt_conv2d_30, opt_conv2d_36, opt_conv2d_42,
+ ))
+ opt_batchnorm2d_44 = self.batchnorm2d_44(opt_concat_43)
+ opt_relu_45 = self.relu_45(opt_batchnorm2d_44)
+ opt_conv2d_46 = self.conv2d_46(opt_relu_45)
+ opt_relu_47 = self.relu_47(opt_conv2d_46)
+ opt_conv2d_48 = self.conv2d_48(opt_relu_47)
+ opt_concat_49 = self.concat_49((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18,
+ opt_conv2d_24, opt_conv2d_30, opt_conv2d_36, opt_conv2d_42, opt_conv2d_48,
+ ))
+ opt_batchnorm2d_50 = self.batchnorm2d_50(opt_concat_49)
+ opt_relu_51 = self.relu_51(opt_batchnorm2d_50)
+ opt_conv2d_52 = self.conv2d_52(opt_relu_51)
+ opt_relu_53 = self.relu_53(opt_conv2d_52)
+ opt_conv2d_54 = self.conv2d_54(opt_relu_53)
+ opt_concat_55 = self.concat_55(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54,
+ ))
+ opt_batchnorm2d_56 = self.batchnorm2d_56(opt_concat_55)
+ opt_relu_57 = self.relu_57(opt_batchnorm2d_56)
+ opt_conv2d_58 = self.conv2d_58(opt_relu_57)
+ opt_relu_59 = self.relu_59(opt_conv2d_58)
+ opt_conv2d_60 = self.conv2d_60(opt_relu_59)
+ opt_concat_61 = self.concat_61(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60,
+ ))
+ opt_batchnorm2d_62 = self.batchnorm2d_62(opt_concat_61)
+ opt_relu_63 = self.relu_63(opt_batchnorm2d_62)
+ opt_conv2d_64 = self.conv2d_64(opt_relu_63)
+ opt_relu_65 = self.relu_65(opt_conv2d_64)
+ opt_conv2d_66 = self.conv2d_66(opt_relu_65)
+ opt_concat_67 = self.concat_67(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66,
+ ))
+ opt_batchnorm2d_68 = self.batchnorm2d_68(opt_concat_67)
+ opt_relu_69 = self.relu_69(opt_batchnorm2d_68)
+ opt_conv2d_70 = self.conv2d_70(opt_relu_69)
+ opt_relu_71 = self.relu_71(opt_conv2d_70)
+ opt_conv2d_72 = self.conv2d_72(opt_relu_71)
+ opt_concat_73 = self.concat_73(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ ))
+ opt_batchnorm2d_74 = self.batchnorm2d_74(opt_concat_73)
+ opt_relu_75 = self.relu_75(opt_batchnorm2d_74)
+ opt_conv2d_76 = self.conv2d_76(opt_relu_75)
+ opt_relu_77 = self.relu_77(opt_conv2d_76)
+ opt_conv2d_78 = self.conv2d_78(opt_relu_77)
+ opt_concat_79 = self.concat_79((opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18,
+ opt_conv2d_24, opt_conv2d_30, opt_conv2d_36, opt_conv2d_42, opt_conv2d_48,
+ opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72, opt_conv2d_78,
+ ))
+ opt_batchnorm2d_80 = self.batchnorm2d_80(opt_concat_79)
+ opt_relu_81 = self.relu_81(opt_batchnorm2d_80)
+ opt_conv2d_82 = self.conv2d_82(opt_relu_81)
+ opt_relu_83 = self.relu_83(opt_conv2d_82)
+ opt_conv2d_84 = self.conv2d_84(opt_relu_83)
+ opt_concat_85 = self.concat_85(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84,
+ ))
+ opt_batchnorm2d_86 = self.batchnorm2d_86(opt_concat_85)
+ opt_relu_87 = self.relu_87(opt_batchnorm2d_86)
+ opt_conv2d_88 = self.conv2d_88(opt_relu_87)
+ opt_relu_89 = self.relu_89(opt_conv2d_88)
+ opt_conv2d_90 = self.conv2d_90(opt_relu_89)
+ opt_concat_91 = self.concat_91(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90,
+ ))
+ opt_batchnorm2d_92 = self.batchnorm2d_92(opt_concat_91)
+ opt_relu_93 = self.relu_93(opt_batchnorm2d_92)
+ opt_conv2d_94 = self.conv2d_94(opt_relu_93)
+ opt_relu_95 = self.relu_95(opt_conv2d_94)
+ opt_conv2d_96 = self.conv2d_96(opt_relu_95)
+ opt_concat_97 = self.concat_97(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96,
+ ))
+ opt_batchnorm2d_98 = self.batchnorm2d_98(opt_concat_97)
+ opt_relu_99 = self.relu_99(opt_batchnorm2d_98)
+ opt_conv2d_100 = self.conv2d_100(opt_relu_99)
+ opt_relu_101 = self.relu_101(opt_conv2d_100)
+ opt_conv2d_102 = self.conv2d_102(opt_relu_101)
+ opt_concat_103 = self.concat_103(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102,
+ ))
+ opt_batchnorm2d_104 = self.batchnorm2d_104(opt_concat_103)
+ opt_relu_105 = self.relu_105(opt_batchnorm2d_104)
+ opt_conv2d_106 = self.conv2d_106(opt_relu_105)
+ opt_relu_107 = self.relu_107(opt_conv2d_106)
+ opt_conv2d_108 = self.conv2d_108(opt_relu_107)
+ opt_concat_109 = self.concat_109(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108,
+ ))
+ opt_batchnorm2d_110 = self.batchnorm2d_110(opt_concat_109)
+ opt_relu_111 = self.relu_111(opt_batchnorm2d_110)
+ opt_conv2d_112 = self.conv2d_112(opt_relu_111)
+ opt_relu_113 = self.relu_113(opt_conv2d_112)
+ opt_conv2d_114 = self.conv2d_114(opt_relu_113)
+ opt_concat_115 = self.concat_115(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ ))
+ opt_batchnorm2d_116 = self.batchnorm2d_116(opt_concat_115)
+ opt_relu_117 = self.relu_117(opt_batchnorm2d_116)
+ opt_conv2d_118 = self.conv2d_118(opt_relu_117)
+ opt_relu_119 = self.relu_119(opt_conv2d_118)
+ opt_conv2d_120 = self.conv2d_120(opt_relu_119)
+ opt_concat_121 = self.concat_121(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120,
+ ))
+ opt_batchnorm2d_122 = self.batchnorm2d_122(opt_concat_121)
+ opt_relu_123 = self.relu_123(opt_batchnorm2d_122)
+ opt_conv2d_124 = self.conv2d_124(opt_relu_123)
+ opt_relu_125 = self.relu_125(opt_conv2d_124)
+ opt_conv2d_126 = self.conv2d_126(opt_relu_125)
+ opt_concat_127 = self.concat_127(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126,
+ ))
+ opt_batchnorm2d_128 = self.batchnorm2d_128(opt_concat_127)
+ opt_relu_129 = self.relu_129(opt_batchnorm2d_128)
+ opt_conv2d_130 = self.conv2d_130(opt_relu_129)
+ opt_relu_131 = self.relu_131(opt_conv2d_130)
+ opt_conv2d_132 = self.conv2d_132(opt_relu_131)
+ opt_concat_133 = self.concat_133(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132,
+ ))
+ opt_batchnorm2d_134 = self.batchnorm2d_134(opt_concat_133)
+ opt_relu_135 = self.relu_135(opt_batchnorm2d_134)
+ opt_conv2d_136 = self.conv2d_136(opt_relu_135)
+ opt_relu_137 = self.relu_137(opt_conv2d_136)
+ opt_conv2d_138 = self.conv2d_138(opt_relu_137)
+ opt_concat_139 = self.concat_139(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138,
+ ))
+ opt_batchnorm2d_140 = self.batchnorm2d_140(opt_concat_139)
+ opt_relu_141 = self.relu_141(opt_batchnorm2d_140)
+ opt_conv2d_142 = self.conv2d_142(opt_relu_141)
+ opt_relu_143 = self.relu_143(opt_conv2d_142)
+ opt_conv2d_144 = self.conv2d_144(opt_relu_143)
+ opt_concat_145 = self.concat_145(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144,
+ ))
+ opt_batchnorm2d_146 = self.batchnorm2d_146(opt_concat_145)
+ opt_relu_147 = self.relu_147(opt_batchnorm2d_146)
+ opt_conv2d_148 = self.conv2d_148(opt_relu_147)
+ opt_relu_149 = self.relu_149(opt_conv2d_148)
+ opt_conv2d_150 = self.conv2d_150(opt_relu_149)
+ opt_concat_151 = self.concat_151(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ ))
+ opt_batchnorm2d_152 = self.batchnorm2d_152(opt_concat_151)
+ opt_relu_153 = self.relu_153(opt_batchnorm2d_152)
+ opt_conv2d_154 = self.conv2d_154(opt_relu_153)
+ opt_relu_155 = self.relu_155(opt_conv2d_154)
+ opt_conv2d_156 = self.conv2d_156(opt_relu_155)
+ opt_concat_157 = self.concat_157(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156,
+ ))
+ opt_batchnorm2d_158 = self.batchnorm2d_158(opt_concat_157)
+ opt_relu_159 = self.relu_159(opt_batchnorm2d_158)
+ opt_conv2d_160 = self.conv2d_160(opt_relu_159)
+ opt_relu_161 = self.relu_161(opt_conv2d_160)
+ opt_conv2d_162 = self.conv2d_162(opt_relu_161)
+ opt_concat_163 = self.concat_163(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156, opt_conv2d_162,
+ ))
+ opt_batchnorm2d_164 = self.batchnorm2d_164(opt_concat_163)
+ opt_relu_165 = self.relu_165(opt_batchnorm2d_164)
+ opt_conv2d_166 = self.conv2d_166(opt_relu_165)
+ opt_relu_167 = self.relu_167(opt_conv2d_166)
+ opt_conv2d_168 = self.conv2d_168(opt_relu_167)
+ opt_concat_169 = self.concat_169(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156, opt_conv2d_162, opt_conv2d_168,
+ ))
+ opt_batchnorm2d_170 = self.batchnorm2d_170(opt_concat_169)
+ opt_relu_171 = self.relu_171(opt_batchnorm2d_170)
+ opt_conv2d_172 = self.conv2d_172(opt_relu_171)
+ opt_relu_173 = self.relu_173(opt_conv2d_172)
+ opt_conv2d_174 = self.conv2d_174(opt_relu_173)
+ opt_concat_175 = self.concat_175(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156, opt_conv2d_162, opt_conv2d_168, opt_conv2d_174,
+ ))
+ opt_batchnorm2d_176 = self.batchnorm2d_176(opt_concat_175)
+ opt_relu_177 = self.relu_177(opt_batchnorm2d_176)
+ opt_conv2d_178 = self.conv2d_178(opt_relu_177)
+ opt_relu_179 = self.relu_179(opt_conv2d_178)
+ opt_conv2d_180 = self.conv2d_180(opt_relu_179)
+ opt_concat_181 = self.concat_181(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156, opt_conv2d_162, opt_conv2d_168, opt_conv2d_174, opt_conv2d_180,
+ ))
+ opt_batchnorm2d_182 = self.batchnorm2d_182(opt_concat_181)
+ opt_relu_183 = self.relu_183(opt_batchnorm2d_182)
+ opt_conv2d_184 = self.conv2d_184(opt_relu_183)
+ opt_relu_185 = self.relu_185(opt_conv2d_184)
+ opt_conv2d_186 = self.conv2d_186(opt_relu_185)
+ opt_concat_187 = self.concat_187(
+ (opt_avgpool2d_0, module0_0_opt, opt_conv2d_6, opt_conv2d_12, opt_conv2d_18, opt_conv2d_24, opt_conv2d_30,
+ opt_conv2d_36, opt_conv2d_42, opt_conv2d_48, opt_conv2d_54, opt_conv2d_60, opt_conv2d_66, opt_conv2d_72,
+ opt_conv2d_78, opt_conv2d_84, opt_conv2d_90, opt_conv2d_96, opt_conv2d_102, opt_conv2d_108, opt_conv2d_114,
+ opt_conv2d_120, opt_conv2d_126, opt_conv2d_132, opt_conv2d_138, opt_conv2d_144, opt_conv2d_150,
+ opt_conv2d_156, opt_conv2d_162, opt_conv2d_168, opt_conv2d_174, opt_conv2d_180, opt_conv2d_186,
+ ))
+ return opt_concat_187
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(7, 7),
+ stride=(2, 2),
+ padding=(3, 3, 3, 3),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.pad_maxpool2d_2 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_2 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module0_0 = Module0(batchnorm2d_1_num_features=64, conv2d_3_in_channels=64)
+ self.concat_9 = P.Concat(axis=1)
+ self.module24_0 = Module24(conv2d_0_in_channels=96, module21_0_batchnorm2d_0_num_features=96)
+ self.concat_15 = P.Concat(axis=1)
+ self.module24_1 = Module24(conv2d_0_in_channels=128, module21_0_batchnorm2d_0_num_features=128)
+ self.concat_21 = P.Concat(axis=1)
+ self.module24_2 = Module24(conv2d_0_in_channels=160, module21_0_batchnorm2d_0_num_features=160)
+ self.concat_27 = P.Concat(axis=1)
+ self.module24_3 = Module24(conv2d_0_in_channels=192, module21_0_batchnorm2d_0_num_features=192)
+ self.concat_33 = P.Concat(axis=1)
+ self.module24_4 = Module24(conv2d_0_in_channels=224, module21_0_batchnorm2d_0_num_features=224)
+ self.concat_39 = P.Concat(axis=1)
+ self.module26_0 = Module26(conv2d_0_in_channels=256,
+ conv2d_0_out_channels=128,
+ module21_0_batchnorm2d_0_num_features=256)
+ self.pad_avgpool2d_43 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_43 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_1 = Module0(batchnorm2d_1_num_features=128, conv2d_3_in_channels=128)
+ self.concat_50 = P.Concat(axis=1)
+ self.module24_5 = Module24(conv2d_0_in_channels=160, module21_0_batchnorm2d_0_num_features=160)
+ self.concat_56 = P.Concat(axis=1)
+ self.module24_6 = Module24(conv2d_0_in_channels=192, module21_0_batchnorm2d_0_num_features=192)
+ self.concat_62 = P.Concat(axis=1)
+ self.module24_7 = Module24(conv2d_0_in_channels=224, module21_0_batchnorm2d_0_num_features=224)
+ self.concat_68 = P.Concat(axis=1)
+ self.module24_8 = Module24(conv2d_0_in_channels=256, module21_0_batchnorm2d_0_num_features=256)
+ self.concat_74 = P.Concat(axis=1)
+ self.module24_9 = Module24(conv2d_0_in_channels=288, module21_0_batchnorm2d_0_num_features=288)
+ self.concat_80 = P.Concat(axis=1)
+ self.module24_10 = Module24(conv2d_0_in_channels=320, module21_0_batchnorm2d_0_num_features=320)
+ self.concat_86 = P.Concat(axis=1)
+ self.module24_11 = Module24(conv2d_0_in_channels=352, module21_0_batchnorm2d_0_num_features=352)
+ self.concat_92 = P.Concat(axis=1)
+ self.module24_12 = Module24(conv2d_0_in_channels=384, module21_0_batchnorm2d_0_num_features=384)
+ self.concat_98 = P.Concat(axis=1)
+ self.module24_13 = Module24(conv2d_0_in_channels=416, module21_0_batchnorm2d_0_num_features=416)
+ self.concat_104 = P.Concat(axis=1)
+ self.module24_14 = Module24(conv2d_0_in_channels=448, module21_0_batchnorm2d_0_num_features=448)
+ self.concat_110 = P.Concat(axis=1)
+ self.module24_15 = Module24(conv2d_0_in_channels=480, module21_0_batchnorm2d_0_num_features=480)
+ self.concat_116 = P.Concat(axis=1)
+ self.module26_1 = Module26(conv2d_0_in_channels=512,
+ conv2d_0_out_channels=256,
+ module21_0_batchnorm2d_0_num_features=512)
+ self.module20_0 = Module20(batchnorm2d_2_num_features=288,
+ conv2d_4_in_channels=288,
+ batchnorm2d_8_num_features=320,
+ conv2d_10_in_channels=320,
+ batchnorm2d_14_num_features=352,
+ conv2d_16_in_channels=352,
+ batchnorm2d_20_num_features=384,
+ conv2d_22_in_channels=384,
+ batchnorm2d_26_num_features=416,
+ conv2d_28_in_channels=416,
+ batchnorm2d_32_num_features=448,
+ conv2d_34_in_channels=448,
+ batchnorm2d_38_num_features=480,
+ conv2d_40_in_channels=480,
+ batchnorm2d_44_num_features=512,
+ conv2d_46_in_channels=512,
+ batchnorm2d_50_num_features=544,
+ conv2d_52_in_channels=544,
+ batchnorm2d_56_num_features=576,
+ conv2d_58_in_channels=576,
+ batchnorm2d_62_num_features=608,
+ conv2d_64_in_channels=608,
+ batchnorm2d_68_num_features=640,
+ conv2d_70_in_channels=640,
+ batchnorm2d_74_num_features=672,
+ conv2d_76_in_channels=672,
+ batchnorm2d_80_num_features=704,
+ conv2d_82_in_channels=704,
+ batchnorm2d_86_num_features=736,
+ conv2d_88_in_channels=736,
+ batchnorm2d_92_num_features=768,
+ conv2d_94_in_channels=768,
+ batchnorm2d_98_num_features=800,
+ conv2d_100_in_channels=800,
+ batchnorm2d_104_num_features=832,
+ conv2d_106_in_channels=832,
+ batchnorm2d_110_num_features=864,
+ conv2d_112_in_channels=864,
+ batchnorm2d_116_num_features=896,
+ conv2d_118_in_channels=896,
+ batchnorm2d_122_num_features=928,
+ conv2d_124_in_channels=928,
+ batchnorm2d_128_num_features=960,
+ conv2d_130_in_channels=960,
+ batchnorm2d_134_num_features=992,
+ conv2d_136_in_channels=992,
+ batchnorm2d_140_num_features=1024,
+ conv2d_142_in_channels=1024,
+ batchnorm2d_146_num_features=1056,
+ conv2d_148_in_channels=1056,
+ batchnorm2d_152_num_features=1088,
+ conv2d_154_in_channels=1088,
+ batchnorm2d_158_num_features=1120,
+ conv2d_160_in_channels=1120,
+ batchnorm2d_164_num_features=1152,
+ conv2d_166_in_channels=1152,
+ batchnorm2d_170_num_features=1184,
+ conv2d_172_in_channels=1184,
+ batchnorm2d_176_num_features=1216,
+ conv2d_178_in_channels=1216,
+ batchnorm2d_182_num_features=1248,
+ conv2d_184_in_channels=1248,
+ module0_0_batchnorm2d_1_num_features=256,
+ module0_0_conv2d_3_in_channels=256)
+ self.module26_2 = Module26(conv2d_0_in_channels=1280,
+ conv2d_0_out_channels=640,
+ module21_0_batchnorm2d_0_num_features=1280)
+ self.module20_1 = Module20(batchnorm2d_2_num_features=672,
+ conv2d_4_in_channels=672,
+ batchnorm2d_8_num_features=704,
+ conv2d_10_in_channels=704,
+ batchnorm2d_14_num_features=736,
+ conv2d_16_in_channels=736,
+ batchnorm2d_20_num_features=768,
+ conv2d_22_in_channels=768,
+ batchnorm2d_26_num_features=800,
+ conv2d_28_in_channels=800,
+ batchnorm2d_32_num_features=832,
+ conv2d_34_in_channels=832,
+ batchnorm2d_38_num_features=864,
+ conv2d_40_in_channels=864,
+ batchnorm2d_44_num_features=896,
+ conv2d_46_in_channels=896,
+ batchnorm2d_50_num_features=928,
+ conv2d_52_in_channels=928,
+ batchnorm2d_56_num_features=960,
+ conv2d_58_in_channels=960,
+ batchnorm2d_62_num_features=992,
+ conv2d_64_in_channels=992,
+ batchnorm2d_68_num_features=1024,
+ conv2d_70_in_channels=1024,
+ batchnorm2d_74_num_features=1056,
+ conv2d_76_in_channels=1056,
+ batchnorm2d_80_num_features=1088,
+ conv2d_82_in_channels=1088,
+ batchnorm2d_86_num_features=1120,
+ conv2d_88_in_channels=1120,
+ batchnorm2d_92_num_features=1152,
+ conv2d_94_in_channels=1152,
+ batchnorm2d_98_num_features=1184,
+ conv2d_100_in_channels=1184,
+ batchnorm2d_104_num_features=1216,
+ conv2d_106_in_channels=1216,
+ batchnorm2d_110_num_features=1248,
+ conv2d_112_in_channels=1248,
+ batchnorm2d_116_num_features=1280,
+ conv2d_118_in_channels=1280,
+ batchnorm2d_122_num_features=1312,
+ conv2d_124_in_channels=1312,
+ batchnorm2d_128_num_features=1344,
+ conv2d_130_in_channels=1344,
+ batchnorm2d_134_num_features=1376,
+ conv2d_136_in_channels=1376,
+ batchnorm2d_140_num_features=1408,
+ conv2d_142_in_channels=1408,
+ batchnorm2d_146_num_features=1440,
+ conv2d_148_in_channels=1440,
+ batchnorm2d_152_num_features=1472,
+ conv2d_154_in_channels=1472,
+ batchnorm2d_158_num_features=1504,
+ conv2d_160_in_channels=1504,
+ batchnorm2d_164_num_features=1536,
+ conv2d_166_in_channels=1536,
+ batchnorm2d_170_num_features=1568,
+ conv2d_172_in_channels=1568,
+ batchnorm2d_176_num_features=1600,
+ conv2d_178_in_channels=1600,
+ batchnorm2d_182_num_features=1632,
+ conv2d_184_in_channels=1632,
+ module0_0_batchnorm2d_1_num_features=640,
+ module0_0_conv2d_3_in_channels=640)
+ self.module21_0 = Module21(batchnorm2d_0_num_features=1664)
+ self.avgpool2d_513 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.flatten_514 = nn.Flatten()
+ self.dense_515 = nn.Dense(in_channels=1664, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_maxpool2d_2 = self.pad_maxpool2d_2(opt_relu_1)
+ opt_maxpool2d_2 = self.maxpool2d_2(opt_maxpool2d_2)
+ module0_0_opt = self.module0_0(opt_maxpool2d_2)
+ opt_concat_9 = self.concat_9((opt_maxpool2d_2, module0_0_opt,))
+ module24_0_opt = self.module24_0(opt_concat_9)
+ opt_concat_15 = self.concat_15((opt_maxpool2d_2, module0_0_opt, module24_0_opt,))
+ module24_1_opt = self.module24_1(opt_concat_15)
+ opt_concat_21 = self.concat_21((opt_maxpool2d_2, module0_0_opt, module24_0_opt, module24_1_opt,))
+ module24_2_opt = self.module24_2(opt_concat_21)
+ opt_concat_27 = self.concat_27((opt_maxpool2d_2, module0_0_opt, module24_0_opt, module24_1_opt, module24_2_opt,
+ ))
+ module24_3_opt = self.module24_3(opt_concat_27)
+ opt_concat_33 = self.concat_33(
+ (opt_maxpool2d_2, module0_0_opt, module24_0_opt, module24_1_opt, module24_2_opt, module24_3_opt,
+ ))
+ module24_4_opt = self.module24_4(opt_concat_33)
+ opt_concat_39 = self.concat_39((opt_maxpool2d_2, module0_0_opt, module24_0_opt, module24_1_opt, module24_2_opt,
+ module24_3_opt, module24_4_opt,
+ ))
+ module26_0_opt = self.module26_0(opt_concat_39)
+ opt_avgpool2d_43 = self.pad_avgpool2d_43(module26_0_opt)
+ opt_avgpool2d_43 = self.avgpool2d_43(opt_avgpool2d_43)
+ module0_1_opt = self.module0_1(opt_avgpool2d_43)
+ opt_concat_50 = self.concat_50((opt_avgpool2d_43, module0_1_opt,))
+ module24_5_opt = self.module24_5(opt_concat_50)
+ opt_concat_56 = self.concat_56((opt_avgpool2d_43, module0_1_opt, module24_5_opt,))
+ module24_6_opt = self.module24_6(opt_concat_56)
+ opt_concat_62 = self.concat_62((opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt,))
+ module24_7_opt = self.module24_7(opt_concat_62)
+ opt_concat_68 = self.concat_68((opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt,
+ ))
+ module24_8_opt = self.module24_8(opt_concat_68)
+ opt_concat_74 = self.concat_74(
+ (opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt, module24_8_opt,
+ ))
+ module24_9_opt = self.module24_9(opt_concat_74)
+ opt_concat_80 = self.concat_80((opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt,
+ module24_8_opt, module24_9_opt,
+ ))
+ module24_10_opt = self.module24_10(opt_concat_80)
+ opt_concat_86 = self.concat_86((opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt,
+ module24_8_opt, module24_9_opt, module24_10_opt,
+ ))
+ module24_11_opt = self.module24_11(opt_concat_86)
+ opt_concat_92 = self.concat_92((opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt,
+ module24_8_opt, module24_9_opt, module24_10_opt, module24_11_opt,
+ ))
+ module24_12_opt = self.module24_12(opt_concat_92)
+ opt_concat_98 = self.concat_98(
+ (opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt, module24_8_opt,
+ module24_9_opt, module24_10_opt, module24_11_opt, module24_12_opt,
+ ))
+ module24_13_opt = self.module24_13(opt_concat_98)
+ opt_concat_104 = self.concat_104(
+ (opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt, module24_8_opt,
+ module24_9_opt, module24_10_opt, module24_11_opt, module24_12_opt, module24_13_opt,
+ ))
+ module24_14_opt = self.module24_14(opt_concat_104)
+ opt_concat_110 = self.concat_110(
+ (opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt, module24_8_opt,
+ module24_9_opt, module24_10_opt, module24_11_opt, module24_12_opt, module24_13_opt, module24_14_opt,
+ ))
+ module24_15_opt = self.module24_15(opt_concat_110)
+ opt_concat_116 = self.concat_116(
+ (opt_avgpool2d_43, module0_1_opt, module24_5_opt, module24_6_opt, module24_7_opt, module24_8_opt,
+ module24_9_opt, module24_10_opt, module24_11_opt, module24_12_opt, module24_13_opt, module24_14_opt,
+ module24_15_opt,
+ ))
+ module26_1_opt = self.module26_1(opt_concat_116)
+ module20_0_opt = self.module20_0(module26_1_opt)
+ module26_2_opt = self.module26_2(module20_0_opt)
+ module20_1_opt = self.module20_1(module26_2_opt)
+ module21_0_opt = self.module21_0(module20_1_opt)
+ opt_avgpool2d_513 = self.avgpool2d_513(module21_0_opt)
+ opt_flatten_514 = self.flatten_514(opt_avgpool2d_513)
+ opt_dense_515 = self.dense_515(opt_flatten_514)
+ return opt_dense_515
diff --git a/research/cvtmodel/densenet/src/densenet201.py b/research/cvtmodel/densenet/src/densenet201.py
new file mode 100644
index 0000000000000000000000000000000000000000..644b3e826b1f4e46cee9203e3467e4fac4eb4a70
--- /dev/null
+++ b/research/cvtmodel/densenet/src/densenet201.py
@@ -0,0 +1,975 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module0(nn.Cell):
+ def __init__(self, batchnorm2d_1_num_features, conv2d_3_in_channels):
+ super(Module0, self).__init__()
+ self.concat_0 = P.Concat(axis=1)
+ self.batchnorm2d_1 = nn.BatchNorm2d(num_features=batchnorm2d_1_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_4 = nn.ReLU()
+ self.conv2d_5 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ opt_concat_0 = self.concat_0((x,))
+ opt_batchnorm2d_1 = self.batchnorm2d_1(opt_concat_0)
+ opt_relu_2 = self.relu_2(opt_batchnorm2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_relu_4 = self.relu_4(opt_conv2d_3)
+ opt_conv2d_5 = self.conv2d_5(opt_relu_4)
+ return opt_conv2d_5
+
+
+class Module1(nn.Cell):
+ def __init__(self, batchnorm2d_0_num_features):
+ super(Module1, self).__init__()
+ self.batchnorm2d_0 = nn.BatchNorm2d(num_features=batchnorm2d_0_num_features,
+ eps=9.999999747378752e-06,
+ momentum=0.8999999761581421)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_batchnorm2d_0 = self.batchnorm2d_0(x)
+ opt_relu_1 = self.relu_1(opt_batchnorm2d_0)
+ return opt_relu_1
+
+
+class Module4(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, module1_0_batchnorm2d_0_num_features):
+ super(Module4, self).__init__()
+ self.module1_0 = Module1(batchnorm2d_0_num_features=module1_0_batchnorm2d_0_num_features)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.conv2d_2 = nn.Conv2d(in_channels=128,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+
+ def construct(self, x):
+ module1_0_opt = self.module1_0(x)
+ opt_conv2d_0 = self.conv2d_0(module1_0_opt)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_conv2d_2 = self.conv2d_2(opt_relu_1)
+ return opt_conv2d_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(7, 7),
+ stride=(2, 2),
+ padding=(3, 3, 3, 3),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.pad_maxpool2d_2 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_2 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module0_0 = Module0(batchnorm2d_1_num_features=64, conv2d_3_in_channels=64)
+ self.concat_9 = P.Concat(axis=1)
+ self.module4_0 = Module4(conv2d_0_in_channels=96, module1_0_batchnorm2d_0_num_features=96)
+ self.concat_15 = P.Concat(axis=1)
+ self.module4_1 = Module4(conv2d_0_in_channels=128, module1_0_batchnorm2d_0_num_features=128)
+ self.concat_21 = P.Concat(axis=1)
+ self.module4_2 = Module4(conv2d_0_in_channels=160, module1_0_batchnorm2d_0_num_features=160)
+ self.concat_27 = P.Concat(axis=1)
+ self.module4_3 = Module4(conv2d_0_in_channels=192, module1_0_batchnorm2d_0_num_features=192)
+ self.concat_33 = P.Concat(axis=1)
+ self.module4_4 = Module4(conv2d_0_in_channels=224, module1_0_batchnorm2d_0_num_features=224)
+ self.concat_39 = P.Concat(axis=1)
+ self.module1_0 = Module1(batchnorm2d_0_num_features=256)
+ self.conv2d_42 = nn.Conv2d(in_channels=256,
+ out_channels=128,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_43 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_43 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_1 = Module0(batchnorm2d_1_num_features=128, conv2d_3_in_channels=128)
+ self.concat_50 = P.Concat(axis=1)
+ self.module4_5 = Module4(conv2d_0_in_channels=160, module1_0_batchnorm2d_0_num_features=160)
+ self.concat_56 = P.Concat(axis=1)
+ self.module4_6 = Module4(conv2d_0_in_channels=192, module1_0_batchnorm2d_0_num_features=192)
+ self.concat_62 = P.Concat(axis=1)
+ self.module4_7 = Module4(conv2d_0_in_channels=224, module1_0_batchnorm2d_0_num_features=224)
+ self.concat_68 = P.Concat(axis=1)
+ self.module4_8 = Module4(conv2d_0_in_channels=256, module1_0_batchnorm2d_0_num_features=256)
+ self.concat_74 = P.Concat(axis=1)
+ self.module4_9 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_80 = P.Concat(axis=1)
+ self.module4_10 = Module4(conv2d_0_in_channels=320, module1_0_batchnorm2d_0_num_features=320)
+ self.concat_86 = P.Concat(axis=1)
+ self.module4_11 = Module4(conv2d_0_in_channels=352, module1_0_batchnorm2d_0_num_features=352)
+ self.concat_92 = P.Concat(axis=1)
+ self.module4_12 = Module4(conv2d_0_in_channels=384, module1_0_batchnorm2d_0_num_features=384)
+ self.concat_98 = P.Concat(axis=1)
+ self.module4_13 = Module4(conv2d_0_in_channels=416, module1_0_batchnorm2d_0_num_features=416)
+ self.concat_104 = P.Concat(axis=1)
+ self.module4_14 = Module4(conv2d_0_in_channels=448, module1_0_batchnorm2d_0_num_features=448)
+ self.concat_110 = P.Concat(axis=1)
+ self.module4_15 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_116 = P.Concat(axis=1)
+ self.module1_1 = Module1(batchnorm2d_0_num_features=512)
+ self.conv2d_119 = nn.Conv2d(in_channels=512,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_120 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_120 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_2 = Module0(batchnorm2d_1_num_features=256, conv2d_3_in_channels=256)
+ self.concat_127 = P.Concat(axis=1)
+ self.module4_16 = Module4(conv2d_0_in_channels=288, module1_0_batchnorm2d_0_num_features=288)
+ self.concat_133 = P.Concat(axis=1)
+ self.module4_17 = Module4(conv2d_0_in_channels=320, module1_0_batchnorm2d_0_num_features=320)
+ self.concat_139 = P.Concat(axis=1)
+ self.module4_18 = Module4(conv2d_0_in_channels=352, module1_0_batchnorm2d_0_num_features=352)
+ self.concat_145 = P.Concat(axis=1)
+ self.module4_19 = Module4(conv2d_0_in_channels=384, module1_0_batchnorm2d_0_num_features=384)
+ self.concat_151 = P.Concat(axis=1)
+ self.module4_20 = Module4(conv2d_0_in_channels=416, module1_0_batchnorm2d_0_num_features=416)
+ self.concat_157 = P.Concat(axis=1)
+ self.module4_21 = Module4(conv2d_0_in_channels=448, module1_0_batchnorm2d_0_num_features=448)
+ self.concat_163 = P.Concat(axis=1)
+ self.module4_22 = Module4(conv2d_0_in_channels=480, module1_0_batchnorm2d_0_num_features=480)
+ self.concat_169 = P.Concat(axis=1)
+ self.module4_23 = Module4(conv2d_0_in_channels=512, module1_0_batchnorm2d_0_num_features=512)
+ self.concat_175 = P.Concat(axis=1)
+ self.module4_24 = Module4(conv2d_0_in_channels=544, module1_0_batchnorm2d_0_num_features=544)
+ self.concat_181 = P.Concat(axis=1)
+ self.module4_25 = Module4(conv2d_0_in_channels=576, module1_0_batchnorm2d_0_num_features=576)
+ self.concat_187 = P.Concat(axis=1)
+ self.module4_26 = Module4(conv2d_0_in_channels=608, module1_0_batchnorm2d_0_num_features=608)
+ self.concat_193 = P.Concat(axis=1)
+ self.module4_27 = Module4(conv2d_0_in_channels=640, module1_0_batchnorm2d_0_num_features=640)
+ self.concat_199 = P.Concat(axis=1)
+ self.module4_28 = Module4(conv2d_0_in_channels=672, module1_0_batchnorm2d_0_num_features=672)
+ self.concat_205 = P.Concat(axis=1)
+ self.module4_29 = Module4(conv2d_0_in_channels=704, module1_0_batchnorm2d_0_num_features=704)
+ self.concat_211 = P.Concat(axis=1)
+ self.module4_30 = Module4(conv2d_0_in_channels=736, module1_0_batchnorm2d_0_num_features=736)
+ self.concat_217 = P.Concat(axis=1)
+ self.module4_31 = Module4(conv2d_0_in_channels=768, module1_0_batchnorm2d_0_num_features=768)
+ self.concat_223 = P.Concat(axis=1)
+ self.module4_32 = Module4(conv2d_0_in_channels=800, module1_0_batchnorm2d_0_num_features=800)
+ self.concat_229 = P.Concat(axis=1)
+ self.module4_33 = Module4(conv2d_0_in_channels=832, module1_0_batchnorm2d_0_num_features=832)
+ self.concat_235 = P.Concat(axis=1)
+ self.module4_34 = Module4(conv2d_0_in_channels=864, module1_0_batchnorm2d_0_num_features=864)
+ self.concat_241 = P.Concat(axis=1)
+ self.module4_35 = Module4(conv2d_0_in_channels=896, module1_0_batchnorm2d_0_num_features=896)
+ self.concat_247 = P.Concat(axis=1)
+ self.module4_36 = Module4(conv2d_0_in_channels=928, module1_0_batchnorm2d_0_num_features=928)
+ self.concat_253 = P.Concat(axis=1)
+ self.module4_37 = Module4(conv2d_0_in_channels=960, module1_0_batchnorm2d_0_num_features=960)
+ self.concat_259 = P.Concat(axis=1)
+ self.module4_38 = Module4(conv2d_0_in_channels=992, module1_0_batchnorm2d_0_num_features=992)
+ self.concat_265 = P.Concat(axis=1)
+ self.module4_39 = Module4(conv2d_0_in_channels=1024, module1_0_batchnorm2d_0_num_features=1024)
+ self.concat_271 = P.Concat(axis=1)
+ self.module4_40 = Module4(conv2d_0_in_channels=1056, module1_0_batchnorm2d_0_num_features=1056)
+ self.concat_277 = P.Concat(axis=1)
+ self.module4_41 = Module4(conv2d_0_in_channels=1088, module1_0_batchnorm2d_0_num_features=1088)
+ self.concat_283 = P.Concat(axis=1)
+ self.module4_42 = Module4(conv2d_0_in_channels=1120, module1_0_batchnorm2d_0_num_features=1120)
+ self.concat_289 = P.Concat(axis=1)
+ self.module4_43 = Module4(conv2d_0_in_channels=1152, module1_0_batchnorm2d_0_num_features=1152)
+ self.concat_295 = P.Concat(axis=1)
+ self.module4_44 = Module4(conv2d_0_in_channels=1184, module1_0_batchnorm2d_0_num_features=1184)
+ self.concat_301 = P.Concat(axis=1)
+ self.module4_45 = Module4(conv2d_0_in_channels=1216, module1_0_batchnorm2d_0_num_features=1216)
+ self.concat_307 = P.Concat(axis=1)
+ self.module4_46 = Module4(conv2d_0_in_channels=1248, module1_0_batchnorm2d_0_num_features=1248)
+ self.concat_313 = P.Concat(axis=1)
+ self.module4_47 = Module4(conv2d_0_in_channels=1280, module1_0_batchnorm2d_0_num_features=1280)
+ self.concat_319 = P.Concat(axis=1)
+ self.module4_48 = Module4(conv2d_0_in_channels=1312, module1_0_batchnorm2d_0_num_features=1312)
+ self.concat_325 = P.Concat(axis=1)
+ self.module4_49 = Module4(conv2d_0_in_channels=1344, module1_0_batchnorm2d_0_num_features=1344)
+ self.concat_331 = P.Concat(axis=1)
+ self.module4_50 = Module4(conv2d_0_in_channels=1376, module1_0_batchnorm2d_0_num_features=1376)
+ self.concat_337 = P.Concat(axis=1)
+ self.module4_51 = Module4(conv2d_0_in_channels=1408, module1_0_batchnorm2d_0_num_features=1408)
+ self.concat_343 = P.Concat(axis=1)
+ self.module4_52 = Module4(conv2d_0_in_channels=1440, module1_0_batchnorm2d_0_num_features=1440)
+ self.concat_349 = P.Concat(axis=1)
+ self.module4_53 = Module4(conv2d_0_in_channels=1472, module1_0_batchnorm2d_0_num_features=1472)
+ self.concat_355 = P.Concat(axis=1)
+ self.module4_54 = Module4(conv2d_0_in_channels=1504, module1_0_batchnorm2d_0_num_features=1504)
+ self.concat_361 = P.Concat(axis=1)
+ self.module4_55 = Module4(conv2d_0_in_channels=1536, module1_0_batchnorm2d_0_num_features=1536)
+ self.concat_367 = P.Concat(axis=1)
+ self.module4_56 = Module4(conv2d_0_in_channels=1568, module1_0_batchnorm2d_0_num_features=1568)
+ self.concat_373 = P.Concat(axis=1)
+ self.module4_57 = Module4(conv2d_0_in_channels=1600, module1_0_batchnorm2d_0_num_features=1600)
+ self.concat_379 = P.Concat(axis=1)
+ self.module4_58 = Module4(conv2d_0_in_channels=1632, module1_0_batchnorm2d_0_num_features=1632)
+ self.concat_385 = P.Concat(axis=1)
+ self.module4_59 = Module4(conv2d_0_in_channels=1664, module1_0_batchnorm2d_0_num_features=1664)
+ self.concat_391 = P.Concat(axis=1)
+ self.module4_60 = Module4(conv2d_0_in_channels=1696, module1_0_batchnorm2d_0_num_features=1696)
+ self.concat_397 = P.Concat(axis=1)
+ self.module4_61 = Module4(conv2d_0_in_channels=1728, module1_0_batchnorm2d_0_num_features=1728)
+ self.concat_403 = P.Concat(axis=1)
+ self.module4_62 = Module4(conv2d_0_in_channels=1760, module1_0_batchnorm2d_0_num_features=1760)
+ self.concat_409 = P.Concat(axis=1)
+ self.module1_2 = Module1(batchnorm2d_0_num_features=1792)
+ self.conv2d_412 = nn.Conv2d(in_channels=1792,
+ out_channels=896,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=False)
+ self.pad_avgpool2d_413 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_413 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.module0_3 = Module0(batchnorm2d_1_num_features=896, conv2d_3_in_channels=896)
+ self.concat_420 = P.Concat(axis=1)
+ self.module4_63 = Module4(conv2d_0_in_channels=928, module1_0_batchnorm2d_0_num_features=928)
+ self.concat_426 = P.Concat(axis=1)
+ self.module4_64 = Module4(conv2d_0_in_channels=960, module1_0_batchnorm2d_0_num_features=960)
+ self.concat_432 = P.Concat(axis=1)
+ self.module4_65 = Module4(conv2d_0_in_channels=992, module1_0_batchnorm2d_0_num_features=992)
+ self.concat_438 = P.Concat(axis=1)
+ self.module4_66 = Module4(conv2d_0_in_channels=1024, module1_0_batchnorm2d_0_num_features=1024)
+ self.concat_444 = P.Concat(axis=1)
+ self.module4_67 = Module4(conv2d_0_in_channels=1056, module1_0_batchnorm2d_0_num_features=1056)
+ self.concat_450 = P.Concat(axis=1)
+ self.module4_68 = Module4(conv2d_0_in_channels=1088, module1_0_batchnorm2d_0_num_features=1088)
+ self.concat_456 = P.Concat(axis=1)
+ self.module4_69 = Module4(conv2d_0_in_channels=1120, module1_0_batchnorm2d_0_num_features=1120)
+ self.concat_462 = P.Concat(axis=1)
+ self.module4_70 = Module4(conv2d_0_in_channels=1152, module1_0_batchnorm2d_0_num_features=1152)
+ self.concat_468 = P.Concat(axis=1)
+ self.module4_71 = Module4(conv2d_0_in_channels=1184, module1_0_batchnorm2d_0_num_features=1184)
+ self.concat_474 = P.Concat(axis=1)
+ self.module4_72 = Module4(conv2d_0_in_channels=1216, module1_0_batchnorm2d_0_num_features=1216)
+ self.concat_480 = P.Concat(axis=1)
+ self.module4_73 = Module4(conv2d_0_in_channels=1248, module1_0_batchnorm2d_0_num_features=1248)
+ self.concat_486 = P.Concat(axis=1)
+ self.module4_74 = Module4(conv2d_0_in_channels=1280, module1_0_batchnorm2d_0_num_features=1280)
+ self.concat_492 = P.Concat(axis=1)
+ self.module4_75 = Module4(conv2d_0_in_channels=1312, module1_0_batchnorm2d_0_num_features=1312)
+ self.concat_498 = P.Concat(axis=1)
+ self.module4_76 = Module4(conv2d_0_in_channels=1344, module1_0_batchnorm2d_0_num_features=1344)
+ self.concat_504 = P.Concat(axis=1)
+ self.module4_77 = Module4(conv2d_0_in_channels=1376, module1_0_batchnorm2d_0_num_features=1376)
+ self.concat_510 = P.Concat(axis=1)
+ self.module4_78 = Module4(conv2d_0_in_channels=1408, module1_0_batchnorm2d_0_num_features=1408)
+ self.concat_516 = P.Concat(axis=1)
+ self.module4_79 = Module4(conv2d_0_in_channels=1440, module1_0_batchnorm2d_0_num_features=1440)
+ self.concat_522 = P.Concat(axis=1)
+ self.module4_80 = Module4(conv2d_0_in_channels=1472, module1_0_batchnorm2d_0_num_features=1472)
+ self.concat_528 = P.Concat(axis=1)
+ self.module4_81 = Module4(conv2d_0_in_channels=1504, module1_0_batchnorm2d_0_num_features=1504)
+ self.concat_534 = P.Concat(axis=1)
+ self.module4_82 = Module4(conv2d_0_in_channels=1536, module1_0_batchnorm2d_0_num_features=1536)
+ self.concat_540 = P.Concat(axis=1)
+ self.module4_83 = Module4(conv2d_0_in_channels=1568, module1_0_batchnorm2d_0_num_features=1568)
+ self.concat_546 = P.Concat(axis=1)
+ self.module4_84 = Module4(conv2d_0_in_channels=1600, module1_0_batchnorm2d_0_num_features=1600)
+ self.concat_552 = P.Concat(axis=1)
+ self.module4_85 = Module4(conv2d_0_in_channels=1632, module1_0_batchnorm2d_0_num_features=1632)
+ self.concat_558 = P.Concat(axis=1)
+ self.module4_86 = Module4(conv2d_0_in_channels=1664, module1_0_batchnorm2d_0_num_features=1664)
+ self.concat_564 = P.Concat(axis=1)
+ self.module4_87 = Module4(conv2d_0_in_channels=1696, module1_0_batchnorm2d_0_num_features=1696)
+ self.concat_570 = P.Concat(axis=1)
+ self.module4_88 = Module4(conv2d_0_in_channels=1728, module1_0_batchnorm2d_0_num_features=1728)
+ self.concat_576 = P.Concat(axis=1)
+ self.module4_89 = Module4(conv2d_0_in_channels=1760, module1_0_batchnorm2d_0_num_features=1760)
+ self.concat_582 = P.Concat(axis=1)
+ self.module4_90 = Module4(conv2d_0_in_channels=1792, module1_0_batchnorm2d_0_num_features=1792)
+ self.concat_588 = P.Concat(axis=1)
+ self.module4_91 = Module4(conv2d_0_in_channels=1824, module1_0_batchnorm2d_0_num_features=1824)
+ self.concat_594 = P.Concat(axis=1)
+ self.module4_92 = Module4(conv2d_0_in_channels=1856, module1_0_batchnorm2d_0_num_features=1856)
+ self.concat_600 = P.Concat(axis=1)
+ self.module4_93 = Module4(conv2d_0_in_channels=1888, module1_0_batchnorm2d_0_num_features=1888)
+ self.concat_606 = P.Concat(axis=1)
+ self.module1_3 = Module1(batchnorm2d_0_num_features=1920)
+ self.avgpool2d_609 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.flatten_610 = nn.Flatten()
+ self.dense_611 = nn.Dense(in_channels=1920, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_maxpool2d_2 = self.pad_maxpool2d_2(opt_relu_1)
+ opt_maxpool2d_2 = self.maxpool2d_2(opt_maxpool2d_2)
+ module0_0_opt = self.module0_0(opt_maxpool2d_2)
+ opt_concat_9 = self.concat_9((opt_maxpool2d_2, module0_0_opt,))
+ module4_0_opt = self.module4_0(opt_concat_9)
+ opt_concat_15 = self.concat_15((opt_maxpool2d_2, module0_0_opt, module4_0_opt,))
+ module4_1_opt = self.module4_1(opt_concat_15)
+ opt_concat_21 = self.concat_21((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt,))
+ module4_2_opt = self.module4_2(opt_concat_21)
+ opt_concat_27 = self.concat_27((opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt,))
+ module4_3_opt = self.module4_3(opt_concat_27)
+ opt_concat_33 = self.concat_33(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt,))
+ module4_4_opt = self.module4_4(opt_concat_33)
+ opt_concat_39 = self.concat_39(
+ (opt_maxpool2d_2, module0_0_opt, module4_0_opt, module4_1_opt, module4_2_opt, module4_3_opt, module4_4_opt,
+ ))
+ module1_0_opt = self.module1_0(opt_concat_39)
+ opt_conv2d_42 = self.conv2d_42(module1_0_opt)
+ opt_avgpool2d_43 = self.pad_avgpool2d_43(opt_conv2d_42)
+ opt_avgpool2d_43 = self.avgpool2d_43(opt_avgpool2d_43)
+ module0_1_opt = self.module0_1(opt_avgpool2d_43)
+ opt_concat_50 = self.concat_50((opt_avgpool2d_43, module0_1_opt,))
+ module4_5_opt = self.module4_5(opt_concat_50)
+ opt_concat_56 = self.concat_56((opt_avgpool2d_43, module0_1_opt, module4_5_opt,))
+ module4_6_opt = self.module4_6(opt_concat_56)
+ opt_concat_62 = self.concat_62((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt,))
+ module4_7_opt = self.module4_7(opt_concat_62)
+ opt_concat_68 = self.concat_68((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,))
+ module4_8_opt = self.module4_8(opt_concat_68)
+ opt_concat_74 = self.concat_74(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt,
+ ))
+ module4_9_opt = self.module4_9(opt_concat_74)
+ opt_concat_80 = self.concat_80(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ ))
+ module4_10_opt = self.module4_10(opt_concat_80)
+ opt_concat_86 = self.concat_86((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt,
+ ))
+ module4_11_opt = self.module4_11(opt_concat_86)
+ opt_concat_92 = self.concat_92((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt,
+ ))
+ module4_12_opt = self.module4_12(opt_concat_92)
+ opt_concat_98 = self.concat_98((opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt,
+ module4_8_opt, module4_9_opt, module4_10_opt, module4_11_opt, module4_12_opt,
+ ))
+ module4_13_opt = self.module4_13(opt_concat_98)
+ opt_concat_104 = self.concat_104(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt,
+ ))
+ module4_14_opt = self.module4_14(opt_concat_104)
+ opt_concat_110 = self.concat_110(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt,
+ ))
+ module4_15_opt = self.module4_15(opt_concat_110)
+ opt_concat_116 = self.concat_116(
+ (opt_avgpool2d_43, module0_1_opt, module4_5_opt, module4_6_opt, module4_7_opt, module4_8_opt, module4_9_opt,
+ module4_10_opt, module4_11_opt, module4_12_opt, module4_13_opt, module4_14_opt, module4_15_opt,
+ ))
+ module1_1_opt = self.module1_1(opt_concat_116)
+ opt_conv2d_119 = self.conv2d_119(module1_1_opt)
+ opt_avgpool2d_120 = self.pad_avgpool2d_120(opt_conv2d_119)
+ opt_avgpool2d_120 = self.avgpool2d_120(opt_avgpool2d_120)
+ module0_2_opt = self.module0_2(opt_avgpool2d_120)
+ opt_concat_127 = self.concat_127((opt_avgpool2d_120, module0_2_opt,))
+ module4_16_opt = self.module4_16(opt_concat_127)
+ opt_concat_133 = self.concat_133((opt_avgpool2d_120, module0_2_opt, module4_16_opt,))
+ module4_17_opt = self.module4_17(opt_concat_133)
+ opt_concat_139 = self.concat_139((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,))
+ module4_18_opt = self.module4_18(opt_concat_139)
+ opt_concat_145 = self.concat_145(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt,
+ ))
+ module4_19_opt = self.module4_19(opt_concat_145)
+ opt_concat_151 = self.concat_151(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ ))
+ module4_20_opt = self.module4_20(opt_concat_151)
+ opt_concat_157 = self.concat_157((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt,
+ ))
+ module4_21_opt = self.module4_21(opt_concat_157)
+ opt_concat_163 = self.concat_163((opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt,
+ module4_18_opt, module4_19_opt, module4_20_opt, module4_21_opt,
+ ))
+ module4_22_opt = self.module4_22(opt_concat_163)
+ opt_concat_169 = self.concat_169(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt,
+ ))
+ module4_23_opt = self.module4_23(opt_concat_169)
+ opt_concat_175 = self.concat_175(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt,
+ ))
+ module4_24_opt = self.module4_24(opt_concat_175)
+ opt_concat_181 = self.concat_181(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt,
+ ))
+ module4_25_opt = self.module4_25(opt_concat_181)
+ opt_concat_187 = self.concat_187(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ ))
+ module4_26_opt = self.module4_26(opt_concat_187)
+ opt_concat_193 = self.concat_193(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt,
+ ))
+ module4_27_opt = self.module4_27(opt_concat_193)
+ opt_concat_199 = self.concat_199(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt,
+ ))
+ module4_28_opt = self.module4_28(opt_concat_199)
+ opt_concat_205 = self.concat_205(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt,
+ ))
+ module4_29_opt = self.module4_29(opt_concat_205)
+ opt_concat_211 = self.concat_211(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt,
+ ))
+ module4_30_opt = self.module4_30(opt_concat_211)
+ opt_concat_217 = self.concat_217(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt,
+ ))
+ module4_31_opt = self.module4_31(opt_concat_217)
+ opt_concat_223 = self.concat_223(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ ))
+ module4_32_opt = self.module4_32(opt_concat_223)
+ opt_concat_229 = self.concat_229(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt,
+ ))
+ module4_33_opt = self.module4_33(opt_concat_229)
+ opt_concat_235 = self.concat_235(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt,
+ ))
+ module4_34_opt = self.module4_34(opt_concat_235)
+ opt_concat_241 = self.concat_241(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt,
+ ))
+ module4_35_opt = self.module4_35(opt_concat_241)
+ opt_concat_247 = self.concat_247(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt,
+ ))
+ module4_36_opt = self.module4_36(opt_concat_247)
+ opt_concat_253 = self.concat_253(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt,
+ ))
+ module4_37_opt = self.module4_37(opt_concat_253)
+ opt_concat_259 = self.concat_259(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ ))
+ module4_38_opt = self.module4_38(opt_concat_259)
+ opt_concat_265 = self.concat_265(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt,
+ ))
+ module4_39_opt = self.module4_39(opt_concat_265)
+ opt_concat_271 = self.concat_271(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt,
+ ))
+ module4_40_opt = self.module4_40(opt_concat_271)
+ opt_concat_277 = self.concat_277(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt,
+ ))
+ module4_41_opt = self.module4_41(opt_concat_277)
+ opt_concat_283 = self.concat_283(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt,
+ ))
+ module4_42_opt = self.module4_42(opt_concat_283)
+ opt_concat_289 = self.concat_289(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt,
+ ))
+ module4_43_opt = self.module4_43(opt_concat_289)
+ opt_concat_295 = self.concat_295(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ ))
+ module4_44_opt = self.module4_44(opt_concat_295)
+ opt_concat_301 = self.concat_301(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt,
+ ))
+ module4_45_opt = self.module4_45(opt_concat_301)
+ opt_concat_307 = self.concat_307(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt,
+ ))
+ module4_46_opt = self.module4_46(opt_concat_307)
+ opt_concat_313 = self.concat_313(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt,
+ ))
+ module4_47_opt = self.module4_47(opt_concat_313)
+ opt_concat_319 = self.concat_319(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt,
+ ))
+ module4_48_opt = self.module4_48(opt_concat_319)
+ opt_concat_325 = self.concat_325(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt,
+ ))
+ module4_49_opt = self.module4_49(opt_concat_325)
+ opt_concat_331 = self.concat_331(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ ))
+ module4_50_opt = self.module4_50(opt_concat_331)
+ opt_concat_337 = self.concat_337(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt,
+ ))
+ module4_51_opt = self.module4_51(opt_concat_337)
+ opt_concat_343 = self.concat_343(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt,
+ ))
+ module4_52_opt = self.module4_52(opt_concat_343)
+ opt_concat_349 = self.concat_349(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt,
+ ))
+ module4_53_opt = self.module4_53(opt_concat_349)
+ opt_concat_355 = self.concat_355(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt,
+ ))
+ module4_54_opt = self.module4_54(opt_concat_355)
+ opt_concat_361 = self.concat_361(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt,
+ ))
+ module4_55_opt = self.module4_55(opt_concat_361)
+ opt_concat_367 = self.concat_367(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ ))
+ module4_56_opt = self.module4_56(opt_concat_367)
+ opt_concat_373 = self.concat_373(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt,
+ ))
+ module4_57_opt = self.module4_57(opt_concat_373)
+ opt_concat_379 = self.concat_379(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt,
+ ))
+ module4_58_opt = self.module4_58(opt_concat_379)
+ opt_concat_385 = self.concat_385(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt, module4_58_opt,
+ ))
+ module4_59_opt = self.module4_59(opt_concat_385)
+ opt_concat_391 = self.concat_391(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt,
+ ))
+ module4_60_opt = self.module4_60(opt_concat_391)
+ opt_concat_397 = self.concat_397(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt,
+ ))
+ module4_61_opt = self.module4_61(opt_concat_397)
+ opt_concat_403 = self.concat_403(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt, module4_61_opt,
+ ))
+ module4_62_opt = self.module4_62(opt_concat_403)
+ opt_concat_409 = self.concat_409(
+ (opt_avgpool2d_120, module0_2_opt, module4_16_opt, module4_17_opt, module4_18_opt, module4_19_opt,
+ module4_20_opt, module4_21_opt, module4_22_opt, module4_23_opt, module4_24_opt, module4_25_opt,
+ module4_26_opt, module4_27_opt, module4_28_opt, module4_29_opt, module4_30_opt, module4_31_opt,
+ module4_32_opt, module4_33_opt, module4_34_opt, module4_35_opt, module4_36_opt, module4_37_opt,
+ module4_38_opt, module4_39_opt, module4_40_opt, module4_41_opt, module4_42_opt, module4_43_opt,
+ module4_44_opt, module4_45_opt, module4_46_opt, module4_47_opt, module4_48_opt, module4_49_opt,
+ module4_50_opt, module4_51_opt, module4_52_opt, module4_53_opt, module4_54_opt, module4_55_opt,
+ module4_56_opt, module4_57_opt, module4_58_opt, module4_59_opt, module4_60_opt, module4_61_opt,
+ module4_62_opt,
+ ))
+ module1_2_opt = self.module1_2(opt_concat_409)
+ opt_conv2d_412 = self.conv2d_412(module1_2_opt)
+ opt_avgpool2d_413 = self.pad_avgpool2d_413(opt_conv2d_412)
+ opt_avgpool2d_413 = self.avgpool2d_413(opt_avgpool2d_413)
+ module0_3_opt = self.module0_3(opt_avgpool2d_413)
+ opt_concat_420 = self.concat_420((opt_avgpool2d_413, module0_3_opt,))
+ module4_63_opt = self.module4_63(opt_concat_420)
+ opt_concat_426 = self.concat_426((opt_avgpool2d_413, module0_3_opt, module4_63_opt,))
+ module4_64_opt = self.module4_64(opt_concat_426)
+ opt_concat_432 = self.concat_432((opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt,))
+ module4_65_opt = self.module4_65(opt_concat_432)
+ opt_concat_438 = self.concat_438(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt,
+ ))
+ module4_66_opt = self.module4_66(opt_concat_438)
+ opt_concat_444 = self.concat_444(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ ))
+ module4_67_opt = self.module4_67(opt_concat_444)
+ opt_concat_450 = self.concat_450((opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt,
+ module4_65_opt, module4_66_opt, module4_67_opt,
+ ))
+ module4_68_opt = self.module4_68(opt_concat_450)
+ opt_concat_456 = self.concat_456((opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt,
+ module4_65_opt, module4_66_opt, module4_67_opt, module4_68_opt,
+ ))
+ module4_69_opt = self.module4_69(opt_concat_456)
+ opt_concat_462 = self.concat_462(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt,
+ ))
+ module4_70_opt = self.module4_70(opt_concat_462)
+ opt_concat_468 = self.concat_468(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt,
+ ))
+ module4_71_opt = self.module4_71(opt_concat_468)
+ opt_concat_474 = self.concat_474(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt,
+ ))
+ module4_72_opt = self.module4_72(opt_concat_474)
+ opt_concat_480 = self.concat_480(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ ))
+ module4_73_opt = self.module4_73(opt_concat_480)
+ opt_concat_486 = self.concat_486(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt,
+ ))
+ module4_74_opt = self.module4_74(opt_concat_486)
+ opt_concat_492 = self.concat_492(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt,
+ ))
+ module4_75_opt = self.module4_75(opt_concat_492)
+ opt_concat_498 = self.concat_498(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt,
+ ))
+ module4_76_opt = self.module4_76(opt_concat_498)
+ opt_concat_504 = self.concat_504(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt,
+ ))
+ module4_77_opt = self.module4_77(opt_concat_504)
+ opt_concat_510 = self.concat_510(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt,
+ ))
+ module4_78_opt = self.module4_78(opt_concat_510)
+ opt_concat_516 = self.concat_516(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ ))
+ module4_79_opt = self.module4_79(opt_concat_516)
+ opt_concat_522 = self.concat_522(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt,
+ ))
+ module4_80_opt = self.module4_80(opt_concat_522)
+ opt_concat_528 = self.concat_528(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt,
+ ))
+ module4_81_opt = self.module4_81(opt_concat_528)
+ opt_concat_534 = self.concat_534(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt,
+ ))
+ module4_82_opt = self.module4_82(opt_concat_534)
+ opt_concat_540 = self.concat_540(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt,
+ ))
+ module4_83_opt = self.module4_83(opt_concat_540)
+ opt_concat_546 = self.concat_546(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt,
+ ))
+ module4_84_opt = self.module4_84(opt_concat_546)
+ opt_concat_552 = self.concat_552(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ ))
+ module4_85_opt = self.module4_85(opt_concat_552)
+ opt_concat_558 = self.concat_558(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt,
+ ))
+ module4_86_opt = self.module4_86(opt_concat_558)
+ opt_concat_564 = self.concat_564(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt,
+ ))
+ module4_87_opt = self.module4_87(opt_concat_564)
+ opt_concat_570 = self.concat_570(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt,
+ ))
+ module4_88_opt = self.module4_88(opt_concat_570)
+ opt_concat_576 = self.concat_576(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt,
+ ))
+ module4_89_opt = self.module4_89(opt_concat_576)
+ opt_concat_582 = self.concat_582(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt, module4_89_opt,
+ ))
+ module4_90_opt = self.module4_90(opt_concat_582)
+ opt_concat_588 = self.concat_588(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt, module4_89_opt, module4_90_opt,
+ ))
+ module4_91_opt = self.module4_91(opt_concat_588)
+ opt_concat_594 = self.concat_594(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt, module4_89_opt, module4_90_opt,
+ module4_91_opt,
+ ))
+ module4_92_opt = self.module4_92(opt_concat_594)
+ opt_concat_600 = self.concat_600(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt, module4_89_opt, module4_90_opt,
+ module4_91_opt, module4_92_opt,
+ ))
+ module4_93_opt = self.module4_93(opt_concat_600)
+ opt_concat_606 = self.concat_606(
+ (opt_avgpool2d_413, module0_3_opt, module4_63_opt, module4_64_opt, module4_65_opt, module4_66_opt,
+ module4_67_opt, module4_68_opt, module4_69_opt, module4_70_opt, module4_71_opt, module4_72_opt,
+ module4_73_opt, module4_74_opt, module4_75_opt, module4_76_opt, module4_77_opt, module4_78_opt,
+ module4_79_opt, module4_80_opt, module4_81_opt, module4_82_opt, module4_83_opt, module4_84_opt,
+ module4_85_opt, module4_86_opt, module4_87_opt, module4_88_opt, module4_89_opt, module4_90_opt,
+ module4_91_opt, module4_92_opt, module4_93_opt,
+ ))
+ module1_3_opt = self.module1_3(opt_concat_606)
+ opt_avgpool2d_609 = self.avgpool2d_609(module1_3_opt)
+ opt_flatten_610 = self.flatten_610(opt_avgpool2d_609)
+ opt_dense_611 = self.dense_611(opt_flatten_610)
+ return opt_dense_611
diff --git a/research/cvtmodel/resnest/README_CN.md b/research/cvtmodel/resnest/README_CN.md
new file mode 100644
index 0000000000000000000000000000000000000000..43b6615a1e833720335f642216eec44ce50d9949
--- /dev/null
+++ b/research/cvtmodel/resnest/README_CN.md
@@ -0,0 +1,128 @@
+# 目录
+
+<!-- TOC -->
+
+- [目录](#目录)
+- [ResNeSt描述](#resnest描述)
+- [模型架构](#模型架构)
+- [数据集](#数据集)
+- [环境要求](#环境要求)
+- [脚本说明](#脚本说明)
+ - [脚本及样例代码](#脚本及样例代码)
+ - [导出过程](#导出过程)
+ - [导出](#导出)
+ - [推理过程](#推理过程)
+ - [推理](#推理)
+- [ModelZoo主页](#modelzoo主页)
+
+<!-- /TOC -->
+
+# ResNeSt描述
+
+ResNeSt是一个基于卷积的神经网络，用于图像分类。有关该模型的描述，可查阅(https://pytorch.org/hub/pytorch_vision_resnest/)。
+本仓库中是基于torch提供的模型文件，使用MindConverter工具转化出Mindspore来ckpt文件，进行全量推理以验证模型文件精度。
+
+# 模型架构
+
+ResNeSt模型支持四种模式：ResNeSt-50, ResNeSt-101，ResNeSt-201, ResNeSt-269。
+
+# 数据集
+
+ResNeSt使用的数据集： ImageNet
+
+数据集的默认配置如下：
+
+- 测试数据集预处理：
+ - 图像的输入尺寸(ResNeSt-50)：224\*224（将图像缩放到256\*256，然后在中央区域裁剪图像）
+ - 图像的输入尺寸(ResNeSt-101)：256\*256（将图像缩放到256\*256，然后在中央区域裁剪图像）
+ - 图像的输入尺寸(ResNeSt-200)：320\*320（将图像缩放到320\*320，然后在中央区域裁剪图像）
+ - 图像的输入尺寸(ResNeSt-269)：416\*416（将图像缩放到416\*416，然后在中央区域裁剪图像）
+ - 根据平均值和标准偏差对输入图像进行归一化
+
+# 环境要求
+
+- 硬件（Ascend/GPU）
+- 准备Ascend或GPU处理器搭建硬件环境。
+- 框架
+- [MindSpore](https://www.mindspore.cn/install)
+- 如需查看详情，请参见如下资源：
+- [MindSpore教程](https://www.mindspore.cn/tutorials/zh-CN/master/index.html)
+- [MindSpore Python API](https://www.mindspore.cn/docs/api/zh-CN/master/index.html)
+
+# 脚本说明
+
+## 脚本及样例代码
+
+```shell
+├── model_zoo
+ ├── README.md // 所有模型的说明
+ ├── ResNeSt
+ ├── README_CN.md // ResNeSt相关说明
+ ├── ascend310_infer // 实现310推理源代码
+ ├── scripts
+ │ ├── run_infer_310.sh // Ascend 310 推理shell脚本
+ ├── src
+ │ ├── resnest50.py // resnest-50模型文件
+ │ ├── resnest101.py // resnest-101模型文件
+ │ ├── resnest200.py // resnest-200模型文件
+ │ ├── resnest269.py // resnest-269模型文件
+ ├── export.py // 导出脚本
+ ├── postprocess.py // 310 推理后处理脚本
+```
+
+## 导出过程
+
+### 导出
+
+```shell
+python export.py --backbone [NET_NAME] --ckpt_path [CKPT_PATH] --device_target [DEVICE_TARGET] --device_id 0 --file_format [EXPORT_FORMAT] --file_name [FILE_NAME]
+```
+
+`backbone` 可选 ["ResNeSt50", "ResNeSt101", "ResNeSt200", "ResNeSt269"]
+`EXPORT_FORMAT` 可选 ["AIR", "MINDIR"]
+
+## 推理过程
+
+### 推理
+
+在推理之前需要先导出模型，AIR模型只能在昇腾910环境上导出，MINDIR可以在任意环境上导出。
+
+```shell
+# 昇腾310 推理
+bash run_infer_310.sh [MINDIR_PATH] [BACKBONE] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+```
+
+-注: Densnet系列网络使用ImageNet数据集,图片的label是将文件夹排序后从0开始编号所得的数字.
+
+推理的结果保存在当前目录下，在acc.log日志文件中可以找到类似以下的结果。
+ResNeSt-50网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=40560, tot=50000, acc=81.12%
+ after allreduce eval: top5_correct=47792, tot=50000, acc=95.58%
+ ```
+
+ResNeSt-101网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=41330, tot=50000, acc=82.66%
+ after allreduce eval: top5_correct=48151, tot=50000, acc=96.30%
+ ```
+
+ResNeSt-200网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=42082, tot=50000, acc=84.16%
+ after allreduce eval: top5_correct=48469, tot=50000, acc=96.94%
+ ```
+
+ResNeSt-269网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=42288, tot=50000, acc=84.58%
+ after allreduce eval: top5_correct=48562, tot=50000, acc=97.12%
+ ```
+
+# ModelZoo主页
+
+ 请浏览官网[主页](https://gitee.com/mindspore/models)。
diff --git a/research/cvtmodel/resnest/ascend310_infer/CMakeLists.txt b/research/cvtmodel/resnest/ascend310_infer/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..ee3c85447340e0449ff2b70ed24f60a17e07b2b6
--- /dev/null
+++ b/research/cvtmodel/resnest/ascend310_infer/CMakeLists.txt
@@ -0,0 +1,14 @@
+cmake_minimum_required(VERSION 3.14.1)
+project(Ascend310Infer)
+add_compile_definitions(_GLIBCXX_USE_CXX11_ABI=0)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O0 -g -std=c++17 -Werror -Wall -fPIE -Wl,--allow-shlib-undefined")
+set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
+option(MINDSPORE_PATH "mindspore install path" "")
+include_directories(${MINDSPORE_PATH})
+include_directories(${MINDSPORE_PATH}/include)
+include_directories(${PROJECT_SRC_ROOT})
+find_library(MS_LIB libmindspore.so ${MINDSPORE_PATH}/lib)
+file(GLOB_RECURSE MD_LIB ${MINDSPORE_PATH}/_c_dataengine*)
+
+add_executable(main src/main.cc src/utils.cc)
+target_link_libraries(main ${MS_LIB} ${MD_LIB} gflags)
diff --git a/research/cvtmodel/resnest/ascend310_infer/build.sh b/research/cvtmodel/resnest/ascend310_infer/build.sh
new file mode 100644
index 0000000000000000000000000000000000000000..770a8851efade7f352039fc8665d307ae1abbb00
--- /dev/null
+++ b/research/cvtmodel/resnest/ascend310_infer/build.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [ ! -d out ]; then
+ mkdir out
+fi
+cd out || exit
+cmake .. \
+ -DMINDSPORE_PATH="`pip show mindspore-ascend | grep Location | awk '{print $2"/mindspore"}' | xargs realpath`"
+make
diff --git a/research/cvtmodel/resnest/ascend310_infer/inc/utils.h b/research/cvtmodel/resnest/ascend310_infer/inc/utils.h
new file mode 100644
index 0000000000000000000000000000000000000000..efebe03a8c1179f5a1f9d5f7ee07e0352a9937c6
--- /dev/null
+++ b/research/cvtmodel/resnest/ascend310_infer/inc/utils.h
@@ -0,0 +1,32 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef MINDSPORE_INFERENCE_UTILS_H_
+#define MINDSPORE_INFERENCE_UTILS_H_
+
+#include <sys/stat.h>
+#include <dirent.h>
+#include <vector>
+#include <string>
+#include <memory>
+#include "include/api/types.h"
+
+std::vector<std::string> GetAllFiles(std::string_view dirName);
+DIR *OpenDir(std::string_view dirName);
+std::string RealPath(std::string_view path);
+mindspore::MSTensor ReadFileToTensor(const std::string &file);
+int WriteResult(const std::string& imageFile, const std::vector<mindspore::MSTensor> &outputs);
+#endif
diff --git a/research/cvtmodel/resnest/ascend310_infer/src/main.cc b/research/cvtmodel/resnest/ascend310_infer/src/main.cc
new file mode 100644
index 0000000000000000000000000000000000000000..5568919e170c25240ea41a234021bd3b717da93b
--- /dev/null
+++ b/research/cvtmodel/resnest/ascend310_infer/src/main.cc
@@ -0,0 +1,174 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <sys/time.h>
+#include <gflags/gflags.h>
+#include <dirent.h>
+#include <iostream>
+#include <string>
+#include <algorithm>
+#include <iosfwd>
+#include <vector>
+#include <fstream>
+
+#include "../inc/utils.h"
+#include "include/dataset/execute.h"
+#include "include/dataset/transforms.h"
+#include "include/dataset/vision.h"
+#include "include/dataset/vision_ascend.h"
+#include "include/api/types.h"
+#include "include/api/model.h"
+#include "include/api/serialization.h"
+#include "include/api/context.h"
+
+using mindspore::Serialization;
+using mindspore::Model;
+using mindspore::Context;
+using mindspore::Status;
+using mindspore::ModelType;
+using mindspore::Graph;
+using mindspore::GraphCell;
+using mindspore::kSuccess;
+using mindspore::MSTensor;
+using mindspore::DataType;
+using mindspore::dataset::Execute;
+using mindspore::dataset::TensorTransform;
+using mindspore::dataset::vision::Decode;
+using mindspore::dataset::vision::Resize;
+using mindspore::dataset::vision::CenterCrop;
+using mindspore::dataset::vision::Normalize;
+using mindspore::dataset::vision::HWC2CHW;
+
+using mindspore::dataset::transforms::TypeCast;
+
+
+DEFINE_string(model_path, "", "model path");
+DEFINE_int32(resize, 256, "resize size");
+DEFINE_int32(crop, 224, "crop size");
+DEFINE_string(dataset, "ImageNet", "dataset: ImageNet or Cifar10");
+DEFINE_string(dataset_path, ".", "dataset path");
+DEFINE_int32(device_id, 0, "device id");
+
+int main(int argc, char **argv) {
+ gflags::ParseCommandLineFlags(&argc, &argv, true);
+ if (RealPath(FLAGS_model_path).empty()) {
+ std::cout << "Invalid model" << std::endl;
+ return 1;
+ }
+
+ std::transform(FLAGS_dataset.begin(), FLAGS_dataset.end(), FLAGS_dataset.begin(), ::tolower);
+
+ auto context = std::make_shared<Context>();
+ auto ascend310_info = std::make_shared<mindspore::Ascend310DeviceInfo>();
+ ascend310_info->SetDeviceID(FLAGS_device_id);
+ context->MutableDeviceInfo().push_back(ascend310_info);
+
+ Graph graph;
+ Status ret = Serialization::Load(FLAGS_model_path, ModelType::kMindIR, &graph);
+ if (ret != kSuccess) {
+ std::cout << "Load model failed." << std::endl;
+ return 1;
+ }
+
+ Model model;
+ ret = model.Build(GraphCell(graph), context);
+ if (ret != kSuccess) {
+ std::cout << "ERROR: Build failed." << std::endl;
+ return 1;
+ }
+
+ std::vector<MSTensor> modelInputs = model.GetInputs();
+
+ auto all_files = GetAllFiles(FLAGS_dataset_path);
+ if (all_files.empty()) {
+ std::cout << "ERROR: no input data." << std::endl;
+ return 1;
+ }
+ float c = static_cast<int>(FLAGS_crop);
+ float mean1 = static_cast<float>(c*0.485);
+ float mean2 = static_cast<float>(c*0.456);
+ float mean3 = static_cast<float>(c*0.406);
+ float std1 = static_cast<float>(c*0.229);
+ float std2 = static_cast<float>(c*0.224);
+ float std3 = static_cast<float>(c*0.225);
+ std::shared_ptr<TensorTransform> decode(new Decode());
+ std::shared_ptr<TensorTransform> resize(new Resize({FLAGS_resize, FLAGS_resize}));
+ std::shared_ptr<TensorTransform> centerCrop(new CenterCrop({FLAGS_crop, FLAGS_crop}));
+ std::shared_ptr<TensorTransform> normImageNet(new Normalize({mean1, mean2, mean3}, {std1, std2, std3}));
+ std::shared_ptr<TensorTransform> normCifar10(new Normalize({127.5, 127.5, 127.5}, {127.5, 127.5, 127.5}));
+ std::shared_ptr<TensorTransform> hwc2chw(new HWC2CHW());
+
+ mindspore::dataset::Execute transformImageNet({decode, resize, centerCrop, normImageNet, hwc2chw});
+ mindspore::dataset::Execute transformCifar10({decode, normCifar10, hwc2chw});
+
+ std::map<double, double> costTime_map;
+
+ size_t size = all_files.size();
+ for (size_t i = 0; i < size; ++i) {
+ struct timeval start;
+ struct timeval end;
+ double startTime_ms;
+ double endTime_ms;
+ std::vector<MSTensor> inputs;
+ std::vector<MSTensor> outputs;
+
+ std::cout << "Start predict input files:" << all_files[i] << std::endl;
+ mindspore::MSTensor image = ReadFileToTensor(all_files[i]);
+
+ if (FLAGS_dataset.compare("cifar10") == 0) {
+ transformCifar10(image, &image);
+ } else if (FLAGS_dataset.compare("imagenet") == 0) {
+ transformImageNet(image, &image);
+ } else {
+ std::cout << "unsupported dataset ...";
+ return 1;
+ }
+
+ inputs.emplace_back(modelInputs[0].Name(), modelInputs[0].DataType(), modelInputs[0].Shape(),
+ image.Data().get(), image.DataSize());
+
+ gettimeofday(&start, NULL);
+ model.Predict(inputs, &outputs);
+ gettimeofday(&end, NULL);
+
+ startTime_ms = (1.0 * start.tv_sec * 1000000 + start.tv_usec) / 1000;
+ endTime_ms = (1.0 * end.tv_sec * 1000000 + end.tv_usec) / 1000;
+ costTime_map.insert(std::pair<double, double>(startTime_ms, endTime_ms));
+ int rst = WriteResult(all_files[i], outputs);
+ if (rst != 0) {
+ std::cout << "write result failed." << std::endl;
+ return rst;
+ }
+ }
+ double average = 0.0;
+ int infer_cnt = 0;
+ char tmpCh[256] = {0};
+ for (auto iter = costTime_map.begin(); iter != costTime_map.end(); iter++) {
+ double diff = 0.0;
+ diff = iter->second - iter->first;
+ average += diff;
+ infer_cnt++;
+ }
+
+ average = average/infer_cnt;
+ std::cout << "NN inference cost average time: "<< average << "ms of infer_count " << infer_cnt << std::endl;
+ std::string file_name = "./time_Result" + std::string("/test_perform_static.txt");
+ std::ofstream file_stream(file_name.c_str(), std::ios::trunc);
+ file_stream << tmpCh;
+ file_stream.close();
+ costTime_map.clear();
+ return 0;
+}
diff --git a/research/cvtmodel/resnest/ascend310_infer/src/utils.cc b/research/cvtmodel/resnest/ascend310_infer/src/utils.cc
new file mode 100644
index 0000000000000000000000000000000000000000..653b1de44962614ac77d44481f8d1a7bde52caaf
--- /dev/null
+++ b/research/cvtmodel/resnest/ascend310_infer/src/utils.cc
@@ -0,0 +1,160 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "inc/utils.h"
+
+#include <fstream>
+#include <algorithm>
+#include <iostream>
+
+using mindspore::MSTensor;
+using mindspore::DataType;
+
+std::vector<std::string> GetAllFiles(std::string_view dirName) {
+ struct dirent *filename;
+ DIR *dir = OpenDir(dirName);
+ if (dir == nullptr) {
+ return {};
+ }
+ std::vector<std::string> dirs;
+ std::vector<std::string> files;
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == "..") {
+ continue;
+ } else if (filename->d_type == DT_DIR) {
+ dirs.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else if (filename->d_type == DT_REG) {
+ files.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else {
+ continue;
+ }
+ }
+
+ for (auto d : dirs) {
+ dir = OpenDir(d);
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == ".." || filename->d_type != DT_REG) {
+ continue;
+ }
+ files.emplace_back(std::string(d) + "/" + filename->d_name);
+ }
+ }
+ std::sort(files.begin(), files.end());
+ for (auto &f : files) {
+ std::cout << "image file: " << f << std::endl;
+ }
+ return files;
+}
+
+int WriteResult(const std::string& imageFile, const std::vector<MSTensor> &outputs) {
+ std::string homePath = "./result_Files";
+ const int INVALID_POINTER = -1;
+ const int ERROR = -2;
+ for (size_t i = 0; i < outputs.size(); ++i) {
+ size_t outputSize;
+ std::shared_ptr<const void> netOutput;
+ netOutput = outputs[i].Data();
+ outputSize = outputs[i].DataSize();
+ int pos = imageFile.rfind('/');
+ std::string fileName(imageFile, pos + 1);
+ fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), '_' + std::to_string(i) + ".bin");
+ std::string outFileName = homePath + "/" + fileName;
+ FILE * outputFile = fopen(outFileName.c_str(), "wb");
+ if (outputFile == nullptr) {
+ std::cout << "open result file " << outFileName << " failed" << std::endl;
+ return INVALID_POINTER;
+ }
+ size_t size = fwrite(netOutput.get(), sizeof(char), outputSize, outputFile);
+ if (size != outputSize) {
+ fclose(outputFile);
+ outputFile = nullptr;
+ std::cout << "write result file " << outFileName << " failed, write size[" << size <<
+ "] is smaller than output size[" << outputSize << "], maybe the disk is full." << std::endl;
+ return ERROR;
+ }
+ fclose(outputFile);
+ outputFile = nullptr;
+ }
+ return 0;
+}
+
+mindspore::MSTensor ReadFileToTensor(const std::string &file) {
+ if (file.empty()) {
+ std::cout << "Pointer file is nullptr" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ std::ifstream ifs(file);
+ if (!ifs.good()) {
+ std::cout << "File: " << file << " is not exist" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ if (!ifs.is_open()) {
+ std::cout << "File: " << file << "open failed" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ ifs.seekg(0, std::ios::end);
+ size_t size = ifs.tellg();
+ mindspore::MSTensor buffer(file, mindspore::DataType::kNumberTypeUInt8, {static_cast<int64_t>(size)}, nullptr, size);
+
+ ifs.seekg(0, std::ios::beg);
+ ifs.read(reinterpret_cast<char *>(buffer.MutableData()), size);
+ ifs.close();
+
+ return buffer;
+}
+
+
+DIR *OpenDir(std::string_view dirName) {
+ if (dirName.empty()) {
+ std::cout << " dirName is null ! " << std::endl;
+ return nullptr;
+ }
+ std::string realPath = RealPath(dirName);
+ struct stat s;
+ lstat(realPath.c_str(), &s);
+ if (!S_ISDIR(s.st_mode)) {
+ std::cout << "dirName is not a valid directory !" << std::endl;
+ return nullptr;
+ }
+ DIR *dir;
+ dir = opendir(realPath.c_str());
+ if (dir == nullptr) {
+ std::cout << "Can not open dir " << dirName << std::endl;
+ return nullptr;
+ }
+ std::cout << "Successfully opened the dir " << dirName << std::endl;
+ return dir;
+}
+
+std::string RealPath(std::string_view path) {
+ char realPathMem[PATH_MAX] = {0};
+ char *realPathRet = nullptr;
+ realPathRet = realpath(path.data(), realPathMem);
+
+ if (realPathRet == nullptr) {
+ std::cout << "File: " << path << " is not exist.";
+ return "";
+ }
+
+ std::string realPath(realPathMem);
+ std::cout << path << " realpath is: " << realPath << std::endl;
+ return realPath;
+}
diff --git a/research/cvtmodel/resnest/export.py b/research/cvtmodel/resnest/export.py
new file mode 100644
index 0000000000000000000000000000000000000000..3f2db5ef22b2c9f3eb4bf403497ce88d8f9137bb
--- /dev/null
+++ b/research/cvtmodel/resnest/export.py
@@ -0,0 +1,63 @@
+# Copyright 2020-2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+"""export checkpoint file into air, onnx, mindir models
+ Suggest run as python export.py --backbone [backbone] --ckpt_path [ckpt_path]
+ --file_name [file_name] --file_format [file format]
+"""
+import argparse
+import numpy as np
+import mindspore as ms
+from mindspore import context, Tensor
+from mindspore.train.serialization import export, load_checkpoint, load_param_into_net
+
+parser = argparse.ArgumentParser(description='post process for 310 inference')
+parser.add_argument("--backbone", type=str, required=True, default="resnest50", help="model backbone")
+parser.add_argument("--ckpt_path", type=str, required=True, help="checkpoint file path")
+parser.add_argument("--file_name", type=str, default="resnest50", help="file name")
+parser.add_argument("--file_format", type=str, default="MINDIR", choices=["MINDIR", "AIR"], help="file format")
+parser.add_argument("--device_target", type=str, default="Ascend", choices=["Ascend", "GPU"], help="device target")
+parser.add_argument("--device_id", type=int, default=0, help="device target")
+args = parser.parse_args()
+
+context.set_context(mode=context.GRAPH_MODE, device_target=args.device_target)
+
+def model_export():
+ '''export main function'''
+ if args.device_target == "Ascend":
+ context.set_context(device_id=args.device_id)
+ if args.backbone == "resnest50":
+ from src.resnest50 import MainModel
+ image_size = 224
+ elif args.backbone == "resnest101":
+ from src.resnest101 import MainModel
+ image_size = 256
+ elif args.backbone == "resnest200":
+ from src.resnest200 import MainModel
+ image_size = 320
+ elif args.backbone == "resnest269":
+ from src.resnest269 import MainModel
+ image_size = 416
+
+ net = MainModel()
+
+ param_dict = load_checkpoint(args.ckpt_path)
+ load_param_into_net(net, param_dict)
+
+ input_arr = Tensor(np.zeros([1, 3, image_size, image_size]), ms.float32)
+ export(net, input_arr, file_name=args.file_name, file_format=args.file_format)
+
+if __name__ == '__main__':
+ model_export()
diff --git a/research/cvtmodel/resnest/postprocess.py b/research/cvtmodel/resnest/postprocess.py
new file mode 100644
index 0000000000000000000000000000000000000000..f80c10276201bd29e53a1efbd2e2a96b3884c3dd
--- /dev/null
+++ b/research/cvtmodel/resnest/postprocess.py
@@ -0,0 +1,82 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+'''post process for 310 inference'''
+import os
+import argparse
+import numpy as np
+
+parser = argparse.ArgumentParser(description='post process for resnest 310 inference')
+parser.add_argument("--dataset", type=str, default="imagenet", help="result file path")
+parser.add_argument("--result_path", type=str, required=True, help="result file path")
+parser.add_argument("--label_file", type=str, required=True, help="label file")
+args = parser.parse_args()
+
+def get_top5_acc(top5_arg, gt_class):
+ sub_count = 0
+ for top5, gt in zip(top5_arg, gt_class):
+ if gt in top5:
+ sub_count += 1
+ return sub_count
+
+def read_label(label_file):
+ '''read lebel file'''
+ f = open(label_file, "r")
+ lines = f.readlines()
+
+ img_label = {}
+ for line in lines:
+ img_id = line.split(":")[0]
+ label = line.split(":")[1]
+ img_label[img_id] = label
+
+ return img_label
+
+def cal_acc(dataset, result_path, label_file):
+ '''acc cal main function'''
+ img_label = read_label(label_file)
+
+ img_tot = 0
+ top1_correct = 0
+ top5_correct = 0
+
+ files = os.listdir(result_path)
+ for file in files:
+ full_file_path = os.path.join(result_path, file)
+ if os.path.isfile(full_file_path):
+ result = np.fromfile(full_file_path, dtype=np.float32).reshape(1, 1000)
+ gt_classes = int(img_label[file[:-6]])
+
+ top1_output = np.argmax(result, (-1))
+ top5_output = np.argsort(result)[:, -5:]
+
+ t1_correct = np.equal(top1_output, gt_classes).sum()
+ top1_correct += t1_correct
+ top5_correct += get_top5_acc(top5_output, [gt_classes])
+ img_tot += 1
+
+ results = [[top1_correct], [top5_correct], [img_tot]]
+
+ results = np.array(results)
+ top1_correct = results[0, 0]
+ top5_correct = results[1, 0]
+ img_tot = results[2, 0]
+ acc1 = 100.0 * top1_correct / img_tot
+ acc5 = 100.0 * top5_correct / img_tot
+ print('after allreduce eval: top1_correct={}, tot={}, acc={:.2f}%'.format(top1_correct, img_tot, acc1))
+ if dataset == 'imagenet':
+ print('after allreduce eval: top5_correct={}, tot={}, acc={:.2f}%'.format(top5_correct, img_tot, acc5))
+
+if __name__ == "__main__":
+ cal_acc(args.dataset, args.result_path, args.label_file)
diff --git a/research/cvtmodel/resnest/scripts/run_infer_310.sh b/research/cvtmodel/resnest/scripts/run_infer_310.sh
new file mode 100644
index 0000000000000000000000000000000000000000..8eb1ec3c64e406e00229f0983c9b2e148078f94e
--- /dev/null
+++ b/research/cvtmodel/resnest/scripts/run_infer_310.sh
@@ -0,0 +1,121 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [[ $# -lt 5 || $# -gt 6 ]]; then
+ echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [BACKBONE] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+ DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
+exit 1
+fi
+
+get_real_path(){
+ if [ "${1:0:1}" == "/" ]; then
+ echo "$1"
+ else
+ echo "$(realpath -m $PWD/$1)"
+ fi
+}
+
+model=$(get_real_path $1)
+backbone=$2
+dataset=$3
+data_path=$(get_real_path $4)
+label_file=$(get_real_path $5)
+
+device_id=0
+
+if [ $# == 6 ]; then
+ device_id=$6
+fi
+
+echo $model
+echo $dataset
+echo $data_path
+echo $label_file
+echo $device_id
+
+export ASCEND_HOME=/usr/local/Ascend/
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+ export PATH=$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export TBE_IMPL_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+ export PYTHONPATH=${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+ export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi
+
+if [ "$backbone" == "resnest50" ]; then
+ resize=256
+ crop=224
+elif [ "$backbone" == "resnest101" ]; then
+ resize=256
+ crop=256
+elif [ "$backbone" == "resnest200" ]; then
+ resize=320
+ crop=320
+elif [ "$backbone" == "resnest269" ]; then
+ resize=416
+ crop=416
+fi
+
+function compile_app()
+{
+ cd ../ascend310_infer || exit
+ if [ -f "Makefile" ]; then
+ make clean
+ fi
+ sh build.sh &> build.log
+
+ if [ $? -ne 0 ]; then
+ echo "compile app code failed"
+ exit 1
+ fi
+ cd - || exit
+}
+
+function infer()
+{
+ if [ -d result_Files ]; then
+ rm -rf ./result_Files
+ fi
+ if [ -d time_Result ]; then
+ rm -rf ./time_Result
+ fi
+ mkdir result_Files
+ mkdir time_Result
+ ../ascend310_infer/out/main --model_path=$model --resize=$resize --crop=$crop --dataset=$dataset --dataset_path=$data_path --device_id=$device_id &> infer.log
+
+ if [ $? -ne 0 ]; then
+ echo "execute inference failed"
+ exit 1
+ fi
+}
+
+function cal_acc()
+{
+ python ../postprocess.py --dataset=$dataset --label_file=$label_file --result_path=result_Files &> acc.log
+ if [ $? -ne 0 ]; then
+ echo "calculate accuracy failed"
+ exit 1
+ fi
+}
+
+compile_app
+infer
+cal_acc
diff --git a/research/cvtmodel/resnest/src/resnest101.py b/research/cvtmodel/resnest/src/resnest101.py
new file mode 100644
index 0000000000000000000000000000000000000000..12b0583ca190ccadc787de653462b57d6bde8007
--- /dev/null
+++ b/research/cvtmodel/resnest/src/resnest101.py
@@ -0,0 +1,1847 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module5(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_0_kernel_size, conv2d_0_padding,
+ conv2d_0_pad_mode, conv2d_0_group):
+ super(Module5, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=conv2d_0_kernel_size,
+ stride=(1, 1),
+ padding=conv2d_0_padding,
+ pad_mode=conv2d_0_pad_mode,
+ dilation=(1, 1),
+ group=conv2d_0_group,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ return opt_relu_1
+
+
+class Module22(nn.Cell):
+ def __init__(self, module5_0_conv2d_0_in_channels, module5_0_conv2d_0_out_channels, module5_0_conv2d_0_kernel_size,
+ module5_0_conv2d_0_padding, module5_0_conv2d_0_pad_mode, module5_0_conv2d_0_group,
+ module5_1_conv2d_0_in_channels, module5_1_conv2d_0_out_channels, module5_1_conv2d_0_kernel_size,
+ module5_1_conv2d_0_padding, module5_1_conv2d_0_pad_mode, module5_1_conv2d_0_group):
+ super(Module22, self).__init__()
+ self.module5_0 = Module5(conv2d_0_in_channels=module5_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_0_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_0_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_0_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_0_conv2d_0_pad_mode,
+ conv2d_0_group=module5_0_conv2d_0_group)
+ self.module5_1 = Module5(conv2d_0_in_channels=module5_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_1_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_1_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_1_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_1_conv2d_0_pad_mode,
+ conv2d_0_group=module5_1_conv2d_0_group)
+
+ def construct(self, x):
+ module5_0_opt = self.module5_0(x)
+ module5_1_opt = self.module5_1(module5_0_opt)
+ return module5_1_opt
+
+
+class Module0(nn.Cell):
+ def __init__(self, avgpool2d_0_kernel_size, conv2d_1_in_channels, conv2d_1_out_channels, conv2d_3_in_channels,
+ conv2d_3_out_channels, reshape_4_shape):
+ super(Module0, self).__init__()
+ self.avgpool2d_0 = nn.AvgPool2d(kernel_size=avgpool2d_0_kernel_size)
+ self.conv2d_1 = nn.Conv2d(in_channels=conv2d_1_in_channels,
+ out_channels=conv2d_1_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=conv2d_3_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.reshape_4 = P.Reshape()
+ self.reshape_4_shape = tuple(reshape_4_shape)
+ self.transpose_5 = P.Transpose()
+
+ def construct(self, x):
+ opt_avgpool2d_0 = self.avgpool2d_0(x)
+ opt_conv2d_1 = self.conv2d_1(opt_avgpool2d_0)
+ opt_relu_2 = self.relu_2(opt_conv2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_reshape_4 = self.reshape_4(opt_conv2d_3, self.reshape_4_shape)
+ opt_transpose_5 = self.transpose_5(opt_reshape_4, (0, 3, 1, 2))
+ return opt_transpose_5
+
+
+class Module20(nn.Cell):
+ def __init__(self, reshape_2_shape, module0_0_avgpool2d_0_kernel_size, module0_0_conv2d_1_in_channels,
+ module0_0_conv2d_1_out_channels, module0_0_conv2d_3_in_channels, module0_0_conv2d_3_out_channels,
+ module0_0_reshape_4_shape):
+ super(Module20, self).__init__()
+ self.module0_0 = Module0(avgpool2d_0_kernel_size=module0_0_avgpool2d_0_kernel_size,
+ conv2d_1_in_channels=module0_0_conv2d_1_in_channels,
+ conv2d_1_out_channels=module0_0_conv2d_1_out_channels,
+ conv2d_3_in_channels=module0_0_conv2d_3_in_channels,
+ conv2d_3_out_channels=module0_0_conv2d_3_out_channels,
+ reshape_4_shape=module0_0_reshape_4_shape)
+ self.softmax_0 = nn.Softmax(axis=3)
+ self.transpose_1 = P.Transpose()
+ self.reshape_2 = P.Reshape()
+ self.reshape_2_shape = tuple(reshape_2_shape)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ opt_softmax_0 = self.softmax_0(module0_0_opt)
+ opt_transpose_1 = self.transpose_1(opt_softmax_0, (0, 3, 2, 1))
+ opt_reshape_2 = self.reshape_2(opt_transpose_1, self.reshape_2_shape)
+ return opt_reshape_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(3, 3),
+ stride=(2, 2),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.module22_0 = Module22(module5_0_conv2d_0_in_channels=64,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(3, 3),
+ module5_0_conv2d_0_padding=(1, 1, 1, 1),
+ module5_0_conv2d_0_pad_mode="pad",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=1)
+ self.pad_maxpool2d_6 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_6 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module22_1 = Module22(module5_0_conv2d_0_in_channels=128,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_13 = P.Split(axis=1, output_num=2)
+ self.add_14_bias = 0.0
+ self.module20_0 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(64, 64),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_25 = P.Reshape()
+ self.reshape_25_shape = tuple([1, 128, 1, 1])
+ self.split_26 = P.Split(axis=1, output_num=2)
+ self.add_29_bias = 0.0
+ self.conv2d_31 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_8 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_8 = nn.AvgPool2d(kernel_size=(1, 1), stride=(1, 1))
+ self.conv2d_10 = nn.Conv2d(in_channels=128,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_33 = nn.ReLU()
+ self.module22_2 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_38 = P.Split(axis=1, output_num=2)
+ self.add_39_bias = 0.0
+ self.module20_1 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(64, 64),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_50 = P.Reshape()
+ self.reshape_50_shape = tuple([1, 128, 1, 1])
+ self.split_51 = P.Split(axis=1, output_num=2)
+ self.add_54_bias = 0.0
+ self.conv2d_56 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_58 = nn.ReLU()
+ self.module22_3 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_63 = P.Split(axis=1, output_num=2)
+ self.add_64_bias = 0.0
+ self.module20_2 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(64, 64),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_75 = P.Reshape()
+ self.reshape_75_shape = tuple([1, 128, 1, 1])
+ self.split_76 = P.Split(axis=1, output_num=2)
+ self.add_79_bias = 0.0
+ self.conv2d_81 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_83 = nn.ReLU()
+ self.module22_4 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_90 = P.Split(axis=1, output_num=2)
+ self.add_91_bias = 0.0
+ self.module20_3 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(64, 64),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_102 = P.Reshape()
+ self.reshape_102_shape = tuple([1, 256, 1, 1])
+ self.split_103 = P.Split(axis=1, output_num=2)
+ self.add_106_bias = 0.0
+ self.pad_108 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_109 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_109 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_110 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_85 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_85 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_87 = nn.Conv2d(in_channels=256,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_112 = nn.ReLU()
+ self.module22_5 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_117 = P.Split(axis=1, output_num=2)
+ self.add_118_bias = 0.0
+ self.module20_4 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(32, 32),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_129 = P.Reshape()
+ self.reshape_129_shape = tuple([1, 256, 1, 1])
+ self.split_130 = P.Split(axis=1, output_num=2)
+ self.add_133_bias = 0.0
+ self.conv2d_135 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_137 = nn.ReLU()
+ self.module22_6 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_142 = P.Split(axis=1, output_num=2)
+ self.add_143_bias = 0.0
+ self.module20_5 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(32, 32),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_154 = P.Reshape()
+ self.reshape_154_shape = tuple([1, 256, 1, 1])
+ self.split_155 = P.Split(axis=1, output_num=2)
+ self.add_158_bias = 0.0
+ self.conv2d_160 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_162 = nn.ReLU()
+ self.module22_7 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_167 = P.Split(axis=1, output_num=2)
+ self.add_168_bias = 0.0
+ self.module20_6 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(32, 32),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_179 = P.Reshape()
+ self.reshape_179_shape = tuple([1, 256, 1, 1])
+ self.split_180 = P.Split(axis=1, output_num=2)
+ self.add_183_bias = 0.0
+ self.conv2d_185 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_187 = nn.ReLU()
+ self.module22_8 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_194 = P.Split(axis=1, output_num=2)
+ self.add_195_bias = 0.0
+ self.module20_7 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(32, 32),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_206 = P.Reshape()
+ self.reshape_206_shape = tuple([1, 512, 1, 1])
+ self.split_207 = P.Split(axis=1, output_num=2)
+ self.add_210_bias = 0.0
+ self.pad_212 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_213 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_213 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_214 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_189 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_189 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_191 = nn.Conv2d(in_channels=512,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_216 = nn.ReLU()
+ self.module22_9 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_221 = P.Split(axis=1, output_num=2)
+ self.add_222_bias = 0.0
+ self.module20_8 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_233 = P.Reshape()
+ self.reshape_233_shape = tuple([1, 512, 1, 1])
+ self.split_234 = P.Split(axis=1, output_num=2)
+ self.add_237_bias = 0.0
+ self.conv2d_239 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_241 = nn.ReLU()
+ self.module22_10 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_246 = P.Split(axis=1, output_num=2)
+ self.add_247_bias = 0.0
+ self.module20_9 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_258 = P.Reshape()
+ self.reshape_258_shape = tuple([1, 512, 1, 1])
+ self.split_259 = P.Split(axis=1, output_num=2)
+ self.add_262_bias = 0.0
+ self.conv2d_264 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_266 = nn.ReLU()
+ self.module22_11 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_271 = P.Split(axis=1, output_num=2)
+ self.add_272_bias = 0.0
+ self.module20_10 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_283 = P.Reshape()
+ self.reshape_283_shape = tuple([1, 512, 1, 1])
+ self.split_284 = P.Split(axis=1, output_num=2)
+ self.add_287_bias = 0.0
+ self.conv2d_289 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_291 = nn.ReLU()
+ self.module22_12 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_296 = P.Split(axis=1, output_num=2)
+ self.add_297_bias = 0.0
+ self.module20_11 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_308 = P.Reshape()
+ self.reshape_308_shape = tuple([1, 512, 1, 1])
+ self.split_309 = P.Split(axis=1, output_num=2)
+ self.add_312_bias = 0.0
+ self.conv2d_314 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_316 = nn.ReLU()
+ self.module22_13 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_321 = P.Split(axis=1, output_num=2)
+ self.add_322_bias = 0.0
+ self.module20_12 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_333 = P.Reshape()
+ self.reshape_333_shape = tuple([1, 512, 1, 1])
+ self.split_334 = P.Split(axis=1, output_num=2)
+ self.add_337_bias = 0.0
+ self.conv2d_339 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_341 = nn.ReLU()
+ self.module22_14 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_346 = P.Split(axis=1, output_num=2)
+ self.add_347_bias = 0.0
+ self.module20_13 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_358 = P.Reshape()
+ self.reshape_358_shape = tuple([1, 512, 1, 1])
+ self.split_359 = P.Split(axis=1, output_num=2)
+ self.add_362_bias = 0.0
+ self.conv2d_364 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_366 = nn.ReLU()
+ self.module22_15 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_371 = P.Split(axis=1, output_num=2)
+ self.add_372_bias = 0.0
+ self.module20_14 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_383 = P.Reshape()
+ self.reshape_383_shape = tuple([1, 512, 1, 1])
+ self.split_384 = P.Split(axis=1, output_num=2)
+ self.add_387_bias = 0.0
+ self.conv2d_389 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_391 = nn.ReLU()
+ self.module22_16 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_396 = P.Split(axis=1, output_num=2)
+ self.add_397_bias = 0.0
+ self.module20_15 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_408 = P.Reshape()
+ self.reshape_408_shape = tuple([1, 512, 1, 1])
+ self.split_409 = P.Split(axis=1, output_num=2)
+ self.add_412_bias = 0.0
+ self.conv2d_414 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_416 = nn.ReLU()
+ self.module22_17 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_421 = P.Split(axis=1, output_num=2)
+ self.add_422_bias = 0.0
+ self.module20_16 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_433 = P.Reshape()
+ self.reshape_433_shape = tuple([1, 512, 1, 1])
+ self.split_434 = P.Split(axis=1, output_num=2)
+ self.add_437_bias = 0.0
+ self.conv2d_439 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_441 = nn.ReLU()
+ self.module22_18 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_446 = P.Split(axis=1, output_num=2)
+ self.add_447_bias = 0.0
+ self.module20_17 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_458 = P.Reshape()
+ self.reshape_458_shape = tuple([1, 512, 1, 1])
+ self.split_459 = P.Split(axis=1, output_num=2)
+ self.add_462_bias = 0.0
+ self.conv2d_464 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_466 = nn.ReLU()
+ self.module22_19 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_471 = P.Split(axis=1, output_num=2)
+ self.add_472_bias = 0.0
+ self.module20_18 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_483 = P.Reshape()
+ self.reshape_483_shape = tuple([1, 512, 1, 1])
+ self.split_484 = P.Split(axis=1, output_num=2)
+ self.add_487_bias = 0.0
+ self.conv2d_489 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_491 = nn.ReLU()
+ self.module22_20 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_496 = P.Split(axis=1, output_num=2)
+ self.add_497_bias = 0.0
+ self.module20_19 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_508 = P.Reshape()
+ self.reshape_508_shape = tuple([1, 512, 1, 1])
+ self.split_509 = P.Split(axis=1, output_num=2)
+ self.add_512_bias = 0.0
+ self.conv2d_514 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_516 = nn.ReLU()
+ self.module22_21 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_521 = P.Split(axis=1, output_num=2)
+ self.add_522_bias = 0.0
+ self.module20_20 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_533 = P.Reshape()
+ self.reshape_533_shape = tuple([1, 512, 1, 1])
+ self.split_534 = P.Split(axis=1, output_num=2)
+ self.add_537_bias = 0.0
+ self.conv2d_539 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_541 = nn.ReLU()
+ self.module22_22 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_546 = P.Split(axis=1, output_num=2)
+ self.add_547_bias = 0.0
+ self.module20_21 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_558 = P.Reshape()
+ self.reshape_558_shape = tuple([1, 512, 1, 1])
+ self.split_559 = P.Split(axis=1, output_num=2)
+ self.add_562_bias = 0.0
+ self.conv2d_564 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_566 = nn.ReLU()
+ self.module22_23 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_571 = P.Split(axis=1, output_num=2)
+ self.add_572_bias = 0.0
+ self.module20_22 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_583 = P.Reshape()
+ self.reshape_583_shape = tuple([1, 512, 1, 1])
+ self.split_584 = P.Split(axis=1, output_num=2)
+ self.add_587_bias = 0.0
+ self.conv2d_589 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_591 = nn.ReLU()
+ self.module22_24 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_596 = P.Split(axis=1, output_num=2)
+ self.add_597_bias = 0.0
+ self.module20_23 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_608 = P.Reshape()
+ self.reshape_608_shape = tuple([1, 512, 1, 1])
+ self.split_609 = P.Split(axis=1, output_num=2)
+ self.add_612_bias = 0.0
+ self.conv2d_614 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_616 = nn.ReLU()
+ self.module22_25 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_621 = P.Split(axis=1, output_num=2)
+ self.add_622_bias = 0.0
+ self.module20_24 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_633 = P.Reshape()
+ self.reshape_633_shape = tuple([1, 512, 1, 1])
+ self.split_634 = P.Split(axis=1, output_num=2)
+ self.add_637_bias = 0.0
+ self.conv2d_639 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_641 = nn.ReLU()
+ self.module22_26 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_646 = P.Split(axis=1, output_num=2)
+ self.add_647_bias = 0.0
+ self.module20_25 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_658 = P.Reshape()
+ self.reshape_658_shape = tuple([1, 512, 1, 1])
+ self.split_659 = P.Split(axis=1, output_num=2)
+ self.add_662_bias = 0.0
+ self.conv2d_664 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_666 = nn.ReLU()
+ self.module22_27 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_671 = P.Split(axis=1, output_num=2)
+ self.add_672_bias = 0.0
+ self.module20_26 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_683 = P.Reshape()
+ self.reshape_683_shape = tuple([1, 512, 1, 1])
+ self.split_684 = P.Split(axis=1, output_num=2)
+ self.add_687_bias = 0.0
+ self.conv2d_689 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_691 = nn.ReLU()
+ self.module22_28 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_696 = P.Split(axis=1, output_num=2)
+ self.add_697_bias = 0.0
+ self.module20_27 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_708 = P.Reshape()
+ self.reshape_708_shape = tuple([1, 512, 1, 1])
+ self.split_709 = P.Split(axis=1, output_num=2)
+ self.add_712_bias = 0.0
+ self.conv2d_714 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_716 = nn.ReLU()
+ self.module22_29 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_721 = P.Split(axis=1, output_num=2)
+ self.add_722_bias = 0.0
+ self.module20_28 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_733 = P.Reshape()
+ self.reshape_733_shape = tuple([1, 512, 1, 1])
+ self.split_734 = P.Split(axis=1, output_num=2)
+ self.add_737_bias = 0.0
+ self.conv2d_739 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_741 = nn.ReLU()
+ self.module22_30 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_746 = P.Split(axis=1, output_num=2)
+ self.add_747_bias = 0.0
+ self.module20_29 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_758 = P.Reshape()
+ self.reshape_758_shape = tuple([1, 512, 1, 1])
+ self.split_759 = P.Split(axis=1, output_num=2)
+ self.add_762_bias = 0.0
+ self.conv2d_764 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_766 = nn.ReLU()
+ self.module22_31 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_773 = P.Split(axis=1, output_num=2)
+ self.add_774_bias = 0.0
+ self.module20_30 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(16, 16),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_785 = P.Reshape()
+ self.reshape_785_shape = tuple([1, 1024, 1, 1])
+ self.split_786 = P.Split(axis=1, output_num=2)
+ self.add_789_bias = 0.0
+ self.pad_791 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_792 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_792 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_793 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_768 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_768 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_770 = nn.Conv2d(in_channels=1024,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_795 = nn.ReLU()
+ self.module22_32 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_800 = P.Split(axis=1, output_num=2)
+ self.add_801_bias = 0.0
+ self.module20_31 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(8, 8),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_812 = P.Reshape()
+ self.reshape_812_shape = tuple([1, 1024, 1, 1])
+ self.split_813 = P.Split(axis=1, output_num=2)
+ self.add_816_bias = 0.0
+ self.conv2d_818 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_820 = nn.ReLU()
+ self.module22_33 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_825 = P.Split(axis=1, output_num=2)
+ self.add_826_bias = 0.0
+ self.module20_32 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(8, 8),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_837 = P.Reshape()
+ self.reshape_837_shape = tuple([1, 1024, 1, 1])
+ self.split_838 = P.Split(axis=1, output_num=2)
+ self.add_841_bias = 0.0
+ self.conv2d_843 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_845 = nn.ReLU()
+ self.avgpool2d_846 = nn.AvgPool2d(kernel_size=(8, 8))
+ self.reshape_847 = P.Reshape()
+ self.reshape_847_shape = tuple([1, 2048])
+ self.flatten_848 = nn.Flatten()
+ self.dense_849 = nn.Dense(in_channels=2048, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ module22_0_opt = self.module22_0(opt_relu_1)
+ opt_maxpool2d_6 = self.pad_maxpool2d_6(module22_0_opt)
+ opt_maxpool2d_6 = self.maxpool2d_6(opt_maxpool2d_6)
+ module22_1_opt = self.module22_1(opt_maxpool2d_6)
+ opt_split_13, opt_split_13_1 = self.split_13(module22_1_opt)
+ opt_add_14 = opt_split_13 + self.add_14_bias
+ opt_add_15 = P.Add()(opt_add_14, opt_split_13_1)
+ module20_0_opt = self.module20_0(opt_add_15)
+ opt_reshape_25 = self.reshape_25(module20_0_opt, self.reshape_25_shape)
+ opt_split_26, opt_split_26_1 = self.split_26(opt_reshape_25)
+ opt_mul_27 = P.Mul()(opt_split_26, opt_split_13)
+ opt_mul_28 = P.Mul()(opt_split_26_1, opt_split_13_1)
+ opt_add_29 = opt_mul_27 + self.add_29_bias
+ opt_add_30 = P.Add()(opt_add_29, opt_mul_28)
+ opt_conv2d_31 = self.conv2d_31(opt_add_30)
+ opt_avgpool2d_8 = self.pad_avgpool2d_8(opt_maxpool2d_6)
+ opt_avgpool2d_8 = self.avgpool2d_8(opt_avgpool2d_8)
+ opt_conv2d_10 = self.conv2d_10(opt_avgpool2d_8)
+ opt_add_32 = P.Add()(opt_conv2d_31, opt_conv2d_10)
+ opt_relu_33 = self.relu_33(opt_add_32)
+ module22_2_opt = self.module22_2(opt_relu_33)
+ opt_split_38, opt_split_38_1 = self.split_38(module22_2_opt)
+ opt_add_39 = opt_split_38 + self.add_39_bias
+ opt_add_40 = P.Add()(opt_add_39, opt_split_38_1)
+ module20_1_opt = self.module20_1(opt_add_40)
+ opt_reshape_50 = self.reshape_50(module20_1_opt, self.reshape_50_shape)
+ opt_split_51, opt_split_51_1 = self.split_51(opt_reshape_50)
+ opt_mul_52 = P.Mul()(opt_split_51, opt_split_38)
+ opt_mul_53 = P.Mul()(opt_split_51_1, opt_split_38_1)
+ opt_add_54 = opt_mul_52 + self.add_54_bias
+ opt_add_55 = P.Add()(opt_add_54, opt_mul_53)
+ opt_conv2d_56 = self.conv2d_56(opt_add_55)
+ opt_add_57 = P.Add()(opt_conv2d_56, opt_relu_33)
+ opt_relu_58 = self.relu_58(opt_add_57)
+ module22_3_opt = self.module22_3(opt_relu_58)
+ opt_split_63, opt_split_63_1 = self.split_63(module22_3_opt)
+ opt_add_64 = opt_split_63 + self.add_64_bias
+ opt_add_65 = P.Add()(opt_add_64, opt_split_63_1)
+ module20_2_opt = self.module20_2(opt_add_65)
+ opt_reshape_75 = self.reshape_75(module20_2_opt, self.reshape_75_shape)
+ opt_split_76, opt_split_76_1 = self.split_76(opt_reshape_75)
+ opt_mul_77 = P.Mul()(opt_split_76, opt_split_63)
+ opt_mul_78 = P.Mul()(opt_split_76_1, opt_split_63_1)
+ opt_add_79 = opt_mul_77 + self.add_79_bias
+ opt_add_80 = P.Add()(opt_add_79, opt_mul_78)
+ opt_conv2d_81 = self.conv2d_81(opt_add_80)
+ opt_add_82 = P.Add()(opt_conv2d_81, opt_relu_58)
+ opt_relu_83 = self.relu_83(opt_add_82)
+ module22_4_opt = self.module22_4(opt_relu_83)
+ opt_split_90, opt_split_90_1 = self.split_90(module22_4_opt)
+ opt_add_91 = opt_split_90 + self.add_91_bias
+ opt_add_92 = P.Add()(opt_add_91, opt_split_90_1)
+ module20_3_opt = self.module20_3(opt_add_92)
+ opt_reshape_102 = self.reshape_102(module20_3_opt, self.reshape_102_shape)
+ opt_split_103, opt_split_103_1 = self.split_103(opt_reshape_102)
+ opt_mul_104 = P.Mul()(opt_split_103, opt_split_90)
+ opt_mul_105 = P.Mul()(opt_split_103_1, opt_split_90_1)
+ opt_add_106 = opt_mul_104 + self.add_106_bias
+ opt_add_107 = P.Add()(opt_add_106, opt_mul_105)
+ opt_pad_108 = self.pad_108(opt_add_107)
+ opt_avgpool2d_109 = self.pad_avgpool2d_109(opt_pad_108)
+ opt_avgpool2d_109 = self.avgpool2d_109(opt_avgpool2d_109)
+ opt_conv2d_110 = self.conv2d_110(opt_avgpool2d_109)
+ opt_avgpool2d_85 = self.pad_avgpool2d_85(opt_relu_83)
+ opt_avgpool2d_85 = self.avgpool2d_85(opt_avgpool2d_85)
+ opt_conv2d_87 = self.conv2d_87(opt_avgpool2d_85)
+ opt_add_111 = P.Add()(opt_conv2d_110, opt_conv2d_87)
+ opt_relu_112 = self.relu_112(opt_add_111)
+ module22_5_opt = self.module22_5(opt_relu_112)
+ opt_split_117, opt_split_117_1 = self.split_117(module22_5_opt)
+ opt_add_118 = opt_split_117 + self.add_118_bias
+ opt_add_119 = P.Add()(opt_add_118, opt_split_117_1)
+ module20_4_opt = self.module20_4(opt_add_119)
+ opt_reshape_129 = self.reshape_129(module20_4_opt, self.reshape_129_shape)
+ opt_split_130, opt_split_130_1 = self.split_130(opt_reshape_129)
+ opt_mul_131 = P.Mul()(opt_split_130, opt_split_117)
+ opt_mul_132 = P.Mul()(opt_split_130_1, opt_split_117_1)
+ opt_add_133 = opt_mul_131 + self.add_133_bias
+ opt_add_134 = P.Add()(opt_add_133, opt_mul_132)
+ opt_conv2d_135 = self.conv2d_135(opt_add_134)
+ opt_add_136 = P.Add()(opt_conv2d_135, opt_relu_112)
+ opt_relu_137 = self.relu_137(opt_add_136)
+ module22_6_opt = self.module22_6(opt_relu_137)
+ opt_split_142, opt_split_142_1 = self.split_142(module22_6_opt)
+ opt_add_143 = opt_split_142 + self.add_143_bias
+ opt_add_144 = P.Add()(opt_add_143, opt_split_142_1)
+ module20_5_opt = self.module20_5(opt_add_144)
+ opt_reshape_154 = self.reshape_154(module20_5_opt, self.reshape_154_shape)
+ opt_split_155, opt_split_155_1 = self.split_155(opt_reshape_154)
+ opt_mul_156 = P.Mul()(opt_split_155, opt_split_142)
+ opt_mul_157 = P.Mul()(opt_split_155_1, opt_split_142_1)
+ opt_add_158 = opt_mul_156 + self.add_158_bias
+ opt_add_159 = P.Add()(opt_add_158, opt_mul_157)
+ opt_conv2d_160 = self.conv2d_160(opt_add_159)
+ opt_add_161 = P.Add()(opt_conv2d_160, opt_relu_137)
+ opt_relu_162 = self.relu_162(opt_add_161)
+ module22_7_opt = self.module22_7(opt_relu_162)
+ opt_split_167, opt_split_167_1 = self.split_167(module22_7_opt)
+ opt_add_168 = opt_split_167 + self.add_168_bias
+ opt_add_169 = P.Add()(opt_add_168, opt_split_167_1)
+ module20_6_opt = self.module20_6(opt_add_169)
+ opt_reshape_179 = self.reshape_179(module20_6_opt, self.reshape_179_shape)
+ opt_split_180, opt_split_180_1 = self.split_180(opt_reshape_179)
+ opt_mul_181 = P.Mul()(opt_split_180, opt_split_167)
+ opt_mul_182 = P.Mul()(opt_split_180_1, opt_split_167_1)
+ opt_add_183 = opt_mul_181 + self.add_183_bias
+ opt_add_184 = P.Add()(opt_add_183, opt_mul_182)
+ opt_conv2d_185 = self.conv2d_185(opt_add_184)
+ opt_add_186 = P.Add()(opt_conv2d_185, opt_relu_162)
+ opt_relu_187 = self.relu_187(opt_add_186)
+ module22_8_opt = self.module22_8(opt_relu_187)
+ opt_split_194, opt_split_194_1 = self.split_194(module22_8_opt)
+ opt_add_195 = opt_split_194 + self.add_195_bias
+ opt_add_196 = P.Add()(opt_add_195, opt_split_194_1)
+ module20_7_opt = self.module20_7(opt_add_196)
+ opt_reshape_206 = self.reshape_206(module20_7_opt, self.reshape_206_shape)
+ opt_split_207, opt_split_207_1 = self.split_207(opt_reshape_206)
+ opt_mul_208 = P.Mul()(opt_split_207, opt_split_194)
+ opt_mul_209 = P.Mul()(opt_split_207_1, opt_split_194_1)
+ opt_add_210 = opt_mul_208 + self.add_210_bias
+ opt_add_211 = P.Add()(opt_add_210, opt_mul_209)
+ opt_pad_212 = self.pad_212(opt_add_211)
+ opt_avgpool2d_213 = self.pad_avgpool2d_213(opt_pad_212)
+ opt_avgpool2d_213 = self.avgpool2d_213(opt_avgpool2d_213)
+ opt_conv2d_214 = self.conv2d_214(opt_avgpool2d_213)
+ opt_avgpool2d_189 = self.pad_avgpool2d_189(opt_relu_187)
+ opt_avgpool2d_189 = self.avgpool2d_189(opt_avgpool2d_189)
+ opt_conv2d_191 = self.conv2d_191(opt_avgpool2d_189)
+ opt_add_215 = P.Add()(opt_conv2d_214, opt_conv2d_191)
+ opt_relu_216 = self.relu_216(opt_add_215)
+ module22_9_opt = self.module22_9(opt_relu_216)
+ opt_split_221, opt_split_221_1 = self.split_221(module22_9_opt)
+ opt_add_222 = opt_split_221 + self.add_222_bias
+ opt_add_223 = P.Add()(opt_add_222, opt_split_221_1)
+ module20_8_opt = self.module20_8(opt_add_223)
+ opt_reshape_233 = self.reshape_233(module20_8_opt, self.reshape_233_shape)
+ opt_split_234, opt_split_234_1 = self.split_234(opt_reshape_233)
+ opt_mul_235 = P.Mul()(opt_split_234, opt_split_221)
+ opt_mul_236 = P.Mul()(opt_split_234_1, opt_split_221_1)
+ opt_add_237 = opt_mul_235 + self.add_237_bias
+ opt_add_238 = P.Add()(opt_add_237, opt_mul_236)
+ opt_conv2d_239 = self.conv2d_239(opt_add_238)
+ opt_add_240 = P.Add()(opt_conv2d_239, opt_relu_216)
+ opt_relu_241 = self.relu_241(opt_add_240)
+ module22_10_opt = self.module22_10(opt_relu_241)
+ opt_split_246, opt_split_246_1 = self.split_246(module22_10_opt)
+ opt_add_247 = opt_split_246 + self.add_247_bias
+ opt_add_248 = P.Add()(opt_add_247, opt_split_246_1)
+ module20_9_opt = self.module20_9(opt_add_248)
+ opt_reshape_258 = self.reshape_258(module20_9_opt, self.reshape_258_shape)
+ opt_split_259, opt_split_259_1 = self.split_259(opt_reshape_258)
+ opt_mul_260 = P.Mul()(opt_split_259, opt_split_246)
+ opt_mul_261 = P.Mul()(opt_split_259_1, opt_split_246_1)
+ opt_add_262 = opt_mul_260 + self.add_262_bias
+ opt_add_263 = P.Add()(opt_add_262, opt_mul_261)
+ opt_conv2d_264 = self.conv2d_264(opt_add_263)
+ opt_add_265 = P.Add()(opt_conv2d_264, opt_relu_241)
+ opt_relu_266 = self.relu_266(opt_add_265)
+ module22_11_opt = self.module22_11(opt_relu_266)
+ opt_split_271, opt_split_271_1 = self.split_271(module22_11_opt)
+ opt_add_272 = opt_split_271 + self.add_272_bias
+ opt_add_273 = P.Add()(opt_add_272, opt_split_271_1)
+ module20_10_opt = self.module20_10(opt_add_273)
+ opt_reshape_283 = self.reshape_283(module20_10_opt, self.reshape_283_shape)
+ opt_split_284, opt_split_284_1 = self.split_284(opt_reshape_283)
+ opt_mul_285 = P.Mul()(opt_split_284, opt_split_271)
+ opt_mul_286 = P.Mul()(opt_split_284_1, opt_split_271_1)
+ opt_add_287 = opt_mul_285 + self.add_287_bias
+ opt_add_288 = P.Add()(opt_add_287, opt_mul_286)
+ opt_conv2d_289 = self.conv2d_289(opt_add_288)
+ opt_add_290 = P.Add()(opt_conv2d_289, opt_relu_266)
+ opt_relu_291 = self.relu_291(opt_add_290)
+ module22_12_opt = self.module22_12(opt_relu_291)
+ opt_split_296, opt_split_296_1 = self.split_296(module22_12_opt)
+ opt_add_297 = opt_split_296 + self.add_297_bias
+ opt_add_298 = P.Add()(opt_add_297, opt_split_296_1)
+ module20_11_opt = self.module20_11(opt_add_298)
+ opt_reshape_308 = self.reshape_308(module20_11_opt, self.reshape_308_shape)
+ opt_split_309, opt_split_309_1 = self.split_309(opt_reshape_308)
+ opt_mul_310 = P.Mul()(opt_split_309, opt_split_296)
+ opt_mul_311 = P.Mul()(opt_split_309_1, opt_split_296_1)
+ opt_add_312 = opt_mul_310 + self.add_312_bias
+ opt_add_313 = P.Add()(opt_add_312, opt_mul_311)
+ opt_conv2d_314 = self.conv2d_314(opt_add_313)
+ opt_add_315 = P.Add()(opt_conv2d_314, opt_relu_291)
+ opt_relu_316 = self.relu_316(opt_add_315)
+ module22_13_opt = self.module22_13(opt_relu_316)
+ opt_split_321, opt_split_321_1 = self.split_321(module22_13_opt)
+ opt_add_322 = opt_split_321 + self.add_322_bias
+ opt_add_323 = P.Add()(opt_add_322, opt_split_321_1)
+ module20_12_opt = self.module20_12(opt_add_323)
+ opt_reshape_333 = self.reshape_333(module20_12_opt, self.reshape_333_shape)
+ opt_split_334, opt_split_334_1 = self.split_334(opt_reshape_333)
+ opt_mul_335 = P.Mul()(opt_split_334, opt_split_321)
+ opt_mul_336 = P.Mul()(opt_split_334_1, opt_split_321_1)
+ opt_add_337 = opt_mul_335 + self.add_337_bias
+ opt_add_338 = P.Add()(opt_add_337, opt_mul_336)
+ opt_conv2d_339 = self.conv2d_339(opt_add_338)
+ opt_add_340 = P.Add()(opt_conv2d_339, opt_relu_316)
+ opt_relu_341 = self.relu_341(opt_add_340)
+ module22_14_opt = self.module22_14(opt_relu_341)
+ opt_split_346, opt_split_346_1 = self.split_346(module22_14_opt)
+ opt_add_347 = opt_split_346 + self.add_347_bias
+ opt_add_348 = P.Add()(opt_add_347, opt_split_346_1)
+ module20_13_opt = self.module20_13(opt_add_348)
+ opt_reshape_358 = self.reshape_358(module20_13_opt, self.reshape_358_shape)
+ opt_split_359, opt_split_359_1 = self.split_359(opt_reshape_358)
+ opt_mul_360 = P.Mul()(opt_split_359, opt_split_346)
+ opt_mul_361 = P.Mul()(opt_split_359_1, opt_split_346_1)
+ opt_add_362 = opt_mul_360 + self.add_362_bias
+ opt_add_363 = P.Add()(opt_add_362, opt_mul_361)
+ opt_conv2d_364 = self.conv2d_364(opt_add_363)
+ opt_add_365 = P.Add()(opt_conv2d_364, opt_relu_341)
+ opt_relu_366 = self.relu_366(opt_add_365)
+ module22_15_opt = self.module22_15(opt_relu_366)
+ opt_split_371, opt_split_371_1 = self.split_371(module22_15_opt)
+ opt_add_372 = opt_split_371 + self.add_372_bias
+ opt_add_373 = P.Add()(opt_add_372, opt_split_371_1)
+ module20_14_opt = self.module20_14(opt_add_373)
+ opt_reshape_383 = self.reshape_383(module20_14_opt, self.reshape_383_shape)
+ opt_split_384, opt_split_384_1 = self.split_384(opt_reshape_383)
+ opt_mul_385 = P.Mul()(opt_split_384, opt_split_371)
+ opt_mul_386 = P.Mul()(opt_split_384_1, opt_split_371_1)
+ opt_add_387 = opt_mul_385 + self.add_387_bias
+ opt_add_388 = P.Add()(opt_add_387, opt_mul_386)
+ opt_conv2d_389 = self.conv2d_389(opt_add_388)
+ opt_add_390 = P.Add()(opt_conv2d_389, opt_relu_366)
+ opt_relu_391 = self.relu_391(opt_add_390)
+ module22_16_opt = self.module22_16(opt_relu_391)
+ opt_split_396, opt_split_396_1 = self.split_396(module22_16_opt)
+ opt_add_397 = opt_split_396 + self.add_397_bias
+ opt_add_398 = P.Add()(opt_add_397, opt_split_396_1)
+ module20_15_opt = self.module20_15(opt_add_398)
+ opt_reshape_408 = self.reshape_408(module20_15_opt, self.reshape_408_shape)
+ opt_split_409, opt_split_409_1 = self.split_409(opt_reshape_408)
+ opt_mul_410 = P.Mul()(opt_split_409, opt_split_396)
+ opt_mul_411 = P.Mul()(opt_split_409_1, opt_split_396_1)
+ opt_add_412 = opt_mul_410 + self.add_412_bias
+ opt_add_413 = P.Add()(opt_add_412, opt_mul_411)
+ opt_conv2d_414 = self.conv2d_414(opt_add_413)
+ opt_add_415 = P.Add()(opt_conv2d_414, opt_relu_391)
+ opt_relu_416 = self.relu_416(opt_add_415)
+ module22_17_opt = self.module22_17(opt_relu_416)
+ opt_split_421, opt_split_421_1 = self.split_421(module22_17_opt)
+ opt_add_422 = opt_split_421 + self.add_422_bias
+ opt_add_423 = P.Add()(opt_add_422, opt_split_421_1)
+ module20_16_opt = self.module20_16(opt_add_423)
+ opt_reshape_433 = self.reshape_433(module20_16_opt, self.reshape_433_shape)
+ opt_split_434, opt_split_434_1 = self.split_434(opt_reshape_433)
+ opt_mul_435 = P.Mul()(opt_split_434, opt_split_421)
+ opt_mul_436 = P.Mul()(opt_split_434_1, opt_split_421_1)
+ opt_add_437 = opt_mul_435 + self.add_437_bias
+ opt_add_438 = P.Add()(opt_add_437, opt_mul_436)
+ opt_conv2d_439 = self.conv2d_439(opt_add_438)
+ opt_add_440 = P.Add()(opt_conv2d_439, opt_relu_416)
+ opt_relu_441 = self.relu_441(opt_add_440)
+ module22_18_opt = self.module22_18(opt_relu_441)
+ opt_split_446, opt_split_446_1 = self.split_446(module22_18_opt)
+ opt_add_447 = opt_split_446 + self.add_447_bias
+ opt_add_448 = P.Add()(opt_add_447, opt_split_446_1)
+ module20_17_opt = self.module20_17(opt_add_448)
+ opt_reshape_458 = self.reshape_458(module20_17_opt, self.reshape_458_shape)
+ opt_split_459, opt_split_459_1 = self.split_459(opt_reshape_458)
+ opt_mul_460 = P.Mul()(opt_split_459, opt_split_446)
+ opt_mul_461 = P.Mul()(opt_split_459_1, opt_split_446_1)
+ opt_add_462 = opt_mul_460 + self.add_462_bias
+ opt_add_463 = P.Add()(opt_add_462, opt_mul_461)
+ opt_conv2d_464 = self.conv2d_464(opt_add_463)
+ opt_add_465 = P.Add()(opt_conv2d_464, opt_relu_441)
+ opt_relu_466 = self.relu_466(opt_add_465)
+ module22_19_opt = self.module22_19(opt_relu_466)
+ opt_split_471, opt_split_471_1 = self.split_471(module22_19_opt)
+ opt_add_472 = opt_split_471 + self.add_472_bias
+ opt_add_473 = P.Add()(opt_add_472, opt_split_471_1)
+ module20_18_opt = self.module20_18(opt_add_473)
+ opt_reshape_483 = self.reshape_483(module20_18_opt, self.reshape_483_shape)
+ opt_split_484, opt_split_484_1 = self.split_484(opt_reshape_483)
+ opt_mul_485 = P.Mul()(opt_split_484, opt_split_471)
+ opt_mul_486 = P.Mul()(opt_split_484_1, opt_split_471_1)
+ opt_add_487 = opt_mul_485 + self.add_487_bias
+ opt_add_488 = P.Add()(opt_add_487, opt_mul_486)
+ opt_conv2d_489 = self.conv2d_489(opt_add_488)
+ opt_add_490 = P.Add()(opt_conv2d_489, opt_relu_466)
+ opt_relu_491 = self.relu_491(opt_add_490)
+ module22_20_opt = self.module22_20(opt_relu_491)
+ opt_split_496, opt_split_496_1 = self.split_496(module22_20_opt)
+ opt_add_497 = opt_split_496 + self.add_497_bias
+ opt_add_498 = P.Add()(opt_add_497, opt_split_496_1)
+ module20_19_opt = self.module20_19(opt_add_498)
+ opt_reshape_508 = self.reshape_508(module20_19_opt, self.reshape_508_shape)
+ opt_split_509, opt_split_509_1 = self.split_509(opt_reshape_508)
+ opt_mul_510 = P.Mul()(opt_split_509, opt_split_496)
+ opt_mul_511 = P.Mul()(opt_split_509_1, opt_split_496_1)
+ opt_add_512 = opt_mul_510 + self.add_512_bias
+ opt_add_513 = P.Add()(opt_add_512, opt_mul_511)
+ opt_conv2d_514 = self.conv2d_514(opt_add_513)
+ opt_add_515 = P.Add()(opt_conv2d_514, opt_relu_491)
+ opt_relu_516 = self.relu_516(opt_add_515)
+ module22_21_opt = self.module22_21(opt_relu_516)
+ opt_split_521, opt_split_521_1 = self.split_521(module22_21_opt)
+ opt_add_522 = opt_split_521 + self.add_522_bias
+ opt_add_523 = P.Add()(opt_add_522, opt_split_521_1)
+ module20_20_opt = self.module20_20(opt_add_523)
+ opt_reshape_533 = self.reshape_533(module20_20_opt, self.reshape_533_shape)
+ opt_split_534, opt_split_534_1 = self.split_534(opt_reshape_533)
+ opt_mul_535 = P.Mul()(opt_split_534, opt_split_521)
+ opt_mul_536 = P.Mul()(opt_split_534_1, opt_split_521_1)
+ opt_add_537 = opt_mul_535 + self.add_537_bias
+ opt_add_538 = P.Add()(opt_add_537, opt_mul_536)
+ opt_conv2d_539 = self.conv2d_539(opt_add_538)
+ opt_add_540 = P.Add()(opt_conv2d_539, opt_relu_516)
+ opt_relu_541 = self.relu_541(opt_add_540)
+ module22_22_opt = self.module22_22(opt_relu_541)
+ opt_split_546, opt_split_546_1 = self.split_546(module22_22_opt)
+ opt_add_547 = opt_split_546 + self.add_547_bias
+ opt_add_548 = P.Add()(opt_add_547, opt_split_546_1)
+ module20_21_opt = self.module20_21(opt_add_548)
+ opt_reshape_558 = self.reshape_558(module20_21_opt, self.reshape_558_shape)
+ opt_split_559, opt_split_559_1 = self.split_559(opt_reshape_558)
+ opt_mul_560 = P.Mul()(opt_split_559, opt_split_546)
+ opt_mul_561 = P.Mul()(opt_split_559_1, opt_split_546_1)
+ opt_add_562 = opt_mul_560 + self.add_562_bias
+ opt_add_563 = P.Add()(opt_add_562, opt_mul_561)
+ opt_conv2d_564 = self.conv2d_564(opt_add_563)
+ opt_add_565 = P.Add()(opt_conv2d_564, opt_relu_541)
+ opt_relu_566 = self.relu_566(opt_add_565)
+ module22_23_opt = self.module22_23(opt_relu_566)
+ opt_split_571, opt_split_571_1 = self.split_571(module22_23_opt)
+ opt_add_572 = opt_split_571 + self.add_572_bias
+ opt_add_573 = P.Add()(opt_add_572, opt_split_571_1)
+ module20_22_opt = self.module20_22(opt_add_573)
+ opt_reshape_583 = self.reshape_583(module20_22_opt, self.reshape_583_shape)
+ opt_split_584, opt_split_584_1 = self.split_584(opt_reshape_583)
+ opt_mul_585 = P.Mul()(opt_split_584, opt_split_571)
+ opt_mul_586 = P.Mul()(opt_split_584_1, opt_split_571_1)
+ opt_add_587 = opt_mul_585 + self.add_587_bias
+ opt_add_588 = P.Add()(opt_add_587, opt_mul_586)
+ opt_conv2d_589 = self.conv2d_589(opt_add_588)
+ opt_add_590 = P.Add()(opt_conv2d_589, opt_relu_566)
+ opt_relu_591 = self.relu_591(opt_add_590)
+ module22_24_opt = self.module22_24(opt_relu_591)
+ opt_split_596, opt_split_596_1 = self.split_596(module22_24_opt)
+ opt_add_597 = opt_split_596 + self.add_597_bias
+ opt_add_598 = P.Add()(opt_add_597, opt_split_596_1)
+ module20_23_opt = self.module20_23(opt_add_598)
+ opt_reshape_608 = self.reshape_608(module20_23_opt, self.reshape_608_shape)
+ opt_split_609, opt_split_609_1 = self.split_609(opt_reshape_608)
+ opt_mul_610 = P.Mul()(opt_split_609, opt_split_596)
+ opt_mul_611 = P.Mul()(opt_split_609_1, opt_split_596_1)
+ opt_add_612 = opt_mul_610 + self.add_612_bias
+ opt_add_613 = P.Add()(opt_add_612, opt_mul_611)
+ opt_conv2d_614 = self.conv2d_614(opt_add_613)
+ opt_add_615 = P.Add()(opt_conv2d_614, opt_relu_591)
+ opt_relu_616 = self.relu_616(opt_add_615)
+ module22_25_opt = self.module22_25(opt_relu_616)
+ opt_split_621, opt_split_621_1 = self.split_621(module22_25_opt)
+ opt_add_622 = opt_split_621 + self.add_622_bias
+ opt_add_623 = P.Add()(opt_add_622, opt_split_621_1)
+ module20_24_opt = self.module20_24(opt_add_623)
+ opt_reshape_633 = self.reshape_633(module20_24_opt, self.reshape_633_shape)
+ opt_split_634, opt_split_634_1 = self.split_634(opt_reshape_633)
+ opt_mul_635 = P.Mul()(opt_split_634, opt_split_621)
+ opt_mul_636 = P.Mul()(opt_split_634_1, opt_split_621_1)
+ opt_add_637 = opt_mul_635 + self.add_637_bias
+ opt_add_638 = P.Add()(opt_add_637, opt_mul_636)
+ opt_conv2d_639 = self.conv2d_639(opt_add_638)
+ opt_add_640 = P.Add()(opt_conv2d_639, opt_relu_616)
+ opt_relu_641 = self.relu_641(opt_add_640)
+ module22_26_opt = self.module22_26(opt_relu_641)
+ opt_split_646, opt_split_646_1 = self.split_646(module22_26_opt)
+ opt_add_647 = opt_split_646 + self.add_647_bias
+ opt_add_648 = P.Add()(opt_add_647, opt_split_646_1)
+ module20_25_opt = self.module20_25(opt_add_648)
+ opt_reshape_658 = self.reshape_658(module20_25_opt, self.reshape_658_shape)
+ opt_split_659, opt_split_659_1 = self.split_659(opt_reshape_658)
+ opt_mul_660 = P.Mul()(opt_split_659, opt_split_646)
+ opt_mul_661 = P.Mul()(opt_split_659_1, opt_split_646_1)
+ opt_add_662 = opt_mul_660 + self.add_662_bias
+ opt_add_663 = P.Add()(opt_add_662, opt_mul_661)
+ opt_conv2d_664 = self.conv2d_664(opt_add_663)
+ opt_add_665 = P.Add()(opt_conv2d_664, opt_relu_641)
+ opt_relu_666 = self.relu_666(opt_add_665)
+ module22_27_opt = self.module22_27(opt_relu_666)
+ opt_split_671, opt_split_671_1 = self.split_671(module22_27_opt)
+ opt_add_672 = opt_split_671 + self.add_672_bias
+ opt_add_673 = P.Add()(opt_add_672, opt_split_671_1)
+ module20_26_opt = self.module20_26(opt_add_673)
+ opt_reshape_683 = self.reshape_683(module20_26_opt, self.reshape_683_shape)
+ opt_split_684, opt_split_684_1 = self.split_684(opt_reshape_683)
+ opt_mul_685 = P.Mul()(opt_split_684, opt_split_671)
+ opt_mul_686 = P.Mul()(opt_split_684_1, opt_split_671_1)
+ opt_add_687 = opt_mul_685 + self.add_687_bias
+ opt_add_688 = P.Add()(opt_add_687, opt_mul_686)
+ opt_conv2d_689 = self.conv2d_689(opt_add_688)
+ opt_add_690 = P.Add()(opt_conv2d_689, opt_relu_666)
+ opt_relu_691 = self.relu_691(opt_add_690)
+ module22_28_opt = self.module22_28(opt_relu_691)
+ opt_split_696, opt_split_696_1 = self.split_696(module22_28_opt)
+ opt_add_697 = opt_split_696 + self.add_697_bias
+ opt_add_698 = P.Add()(opt_add_697, opt_split_696_1)
+ module20_27_opt = self.module20_27(opt_add_698)
+ opt_reshape_708 = self.reshape_708(module20_27_opt, self.reshape_708_shape)
+ opt_split_709, opt_split_709_1 = self.split_709(opt_reshape_708)
+ opt_mul_710 = P.Mul()(opt_split_709, opt_split_696)
+ opt_mul_711 = P.Mul()(opt_split_709_1, opt_split_696_1)
+ opt_add_712 = opt_mul_710 + self.add_712_bias
+ opt_add_713 = P.Add()(opt_add_712, opt_mul_711)
+ opt_conv2d_714 = self.conv2d_714(opt_add_713)
+ opt_add_715 = P.Add()(opt_conv2d_714, opt_relu_691)
+ opt_relu_716 = self.relu_716(opt_add_715)
+ module22_29_opt = self.module22_29(opt_relu_716)
+ opt_split_721, opt_split_721_1 = self.split_721(module22_29_opt)
+ opt_add_722 = opt_split_721 + self.add_722_bias
+ opt_add_723 = P.Add()(opt_add_722, opt_split_721_1)
+ module20_28_opt = self.module20_28(opt_add_723)
+ opt_reshape_733 = self.reshape_733(module20_28_opt, self.reshape_733_shape)
+ opt_split_734, opt_split_734_1 = self.split_734(opt_reshape_733)
+ opt_mul_735 = P.Mul()(opt_split_734, opt_split_721)
+ opt_mul_736 = P.Mul()(opt_split_734_1, opt_split_721_1)
+ opt_add_737 = opt_mul_735 + self.add_737_bias
+ opt_add_738 = P.Add()(opt_add_737, opt_mul_736)
+ opt_conv2d_739 = self.conv2d_739(opt_add_738)
+ opt_add_740 = P.Add()(opt_conv2d_739, opt_relu_716)
+ opt_relu_741 = self.relu_741(opt_add_740)
+ module22_30_opt = self.module22_30(opt_relu_741)
+ opt_split_746, opt_split_746_1 = self.split_746(module22_30_opt)
+ opt_add_747 = opt_split_746 + self.add_747_bias
+ opt_add_748 = P.Add()(opt_add_747, opt_split_746_1)
+ module20_29_opt = self.module20_29(opt_add_748)
+ opt_reshape_758 = self.reshape_758(module20_29_opt, self.reshape_758_shape)
+ opt_split_759, opt_split_759_1 = self.split_759(opt_reshape_758)
+ opt_mul_760 = P.Mul()(opt_split_759, opt_split_746)
+ opt_mul_761 = P.Mul()(opt_split_759_1, opt_split_746_1)
+ opt_add_762 = opt_mul_760 + self.add_762_bias
+ opt_add_763 = P.Add()(opt_add_762, opt_mul_761)
+ opt_conv2d_764 = self.conv2d_764(opt_add_763)
+ opt_add_765 = P.Add()(opt_conv2d_764, opt_relu_741)
+ opt_relu_766 = self.relu_766(opt_add_765)
+ module22_31_opt = self.module22_31(opt_relu_766)
+ opt_split_773, opt_split_773_1 = self.split_773(module22_31_opt)
+ opt_add_774 = opt_split_773 + self.add_774_bias
+ opt_add_775 = P.Add()(opt_add_774, opt_split_773_1)
+ module20_30_opt = self.module20_30(opt_add_775)
+ opt_reshape_785 = self.reshape_785(module20_30_opt, self.reshape_785_shape)
+ opt_split_786, opt_split_786_1 = self.split_786(opt_reshape_785)
+ opt_mul_787 = P.Mul()(opt_split_786, opt_split_773)
+ opt_mul_788 = P.Mul()(opt_split_786_1, opt_split_773_1)
+ opt_add_789 = opt_mul_787 + self.add_789_bias
+ opt_add_790 = P.Add()(opt_add_789, opt_mul_788)
+ opt_pad_791 = self.pad_791(opt_add_790)
+ opt_avgpool2d_792 = self.pad_avgpool2d_792(opt_pad_791)
+ opt_avgpool2d_792 = self.avgpool2d_792(opt_avgpool2d_792)
+ opt_conv2d_793 = self.conv2d_793(opt_avgpool2d_792)
+ opt_avgpool2d_768 = self.pad_avgpool2d_768(opt_relu_766)
+ opt_avgpool2d_768 = self.avgpool2d_768(opt_avgpool2d_768)
+ opt_conv2d_770 = self.conv2d_770(opt_avgpool2d_768)
+ opt_add_794 = P.Add()(opt_conv2d_793, opt_conv2d_770)
+ opt_relu_795 = self.relu_795(opt_add_794)
+ module22_32_opt = self.module22_32(opt_relu_795)
+ opt_split_800, opt_split_800_1 = self.split_800(module22_32_opt)
+ opt_add_801 = opt_split_800 + self.add_801_bias
+ opt_add_802 = P.Add()(opt_add_801, opt_split_800_1)
+ module20_31_opt = self.module20_31(opt_add_802)
+ opt_reshape_812 = self.reshape_812(module20_31_opt, self.reshape_812_shape)
+ opt_split_813, opt_split_813_1 = self.split_813(opt_reshape_812)
+ opt_mul_814 = P.Mul()(opt_split_813, opt_split_800)
+ opt_mul_815 = P.Mul()(opt_split_813_1, opt_split_800_1)
+ opt_add_816 = opt_mul_814 + self.add_816_bias
+ opt_add_817 = P.Add()(opt_add_816, opt_mul_815)
+ opt_conv2d_818 = self.conv2d_818(opt_add_817)
+ opt_add_819 = P.Add()(opt_conv2d_818, opt_relu_795)
+ opt_relu_820 = self.relu_820(opt_add_819)
+ module22_33_opt = self.module22_33(opt_relu_820)
+ opt_split_825, opt_split_825_1 = self.split_825(module22_33_opt)
+ opt_add_826 = opt_split_825 + self.add_826_bias
+ opt_add_827 = P.Add()(opt_add_826, opt_split_825_1)
+ module20_32_opt = self.module20_32(opt_add_827)
+ opt_reshape_837 = self.reshape_837(module20_32_opt, self.reshape_837_shape)
+ opt_split_838, opt_split_838_1 = self.split_838(opt_reshape_837)
+ opt_mul_839 = P.Mul()(opt_split_838, opt_split_825)
+ opt_mul_840 = P.Mul()(opt_split_838_1, opt_split_825_1)
+ opt_add_841 = opt_mul_839 + self.add_841_bias
+ opt_add_842 = P.Add()(opt_add_841, opt_mul_840)
+ opt_conv2d_843 = self.conv2d_843(opt_add_842)
+ opt_add_844 = P.Add()(opt_conv2d_843, opt_relu_820)
+ opt_relu_845 = self.relu_845(opt_add_844)
+ opt_avgpool2d_846 = self.avgpool2d_846(opt_relu_845)
+ opt_reshape_847 = self.reshape_847(opt_avgpool2d_846, self.reshape_847_shape)
+ opt_flatten_848 = self.flatten_848(opt_reshape_847)
+ opt_dense_849 = self.dense_849(opt_flatten_848)
+ return opt_dense_849
diff --git a/research/cvtmodel/resnest/src/resnest200.py b/research/cvtmodel/resnest/src/resnest200.py
new file mode 100644
index 0000000000000000000000000000000000000000..75aba8c4c4cb304d2fcfafa371be4cdcd22c9d0b
--- /dev/null
+++ b/research/cvtmodel/resnest/src/resnest200.py
@@ -0,0 +1,3464 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module5(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_0_kernel_size, conv2d_0_padding,
+ conv2d_0_pad_mode, conv2d_0_group):
+ super(Module5, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=conv2d_0_kernel_size,
+ stride=(1, 1),
+ padding=conv2d_0_padding,
+ pad_mode=conv2d_0_pad_mode,
+ dilation=(1, 1),
+ group=conv2d_0_group,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ return opt_relu_1
+
+
+class Module22(nn.Cell):
+ def __init__(self, module5_0_conv2d_0_in_channels, module5_0_conv2d_0_out_channels, module5_0_conv2d_0_kernel_size,
+ module5_0_conv2d_0_padding, module5_0_conv2d_0_pad_mode, module5_0_conv2d_0_group,
+ module5_1_conv2d_0_in_channels, module5_1_conv2d_0_out_channels, module5_1_conv2d_0_kernel_size,
+ module5_1_conv2d_0_padding, module5_1_conv2d_0_pad_mode, module5_1_conv2d_0_group):
+ super(Module22, self).__init__()
+ self.module5_0 = Module5(conv2d_0_in_channels=module5_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_0_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_0_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_0_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_0_conv2d_0_pad_mode,
+ conv2d_0_group=module5_0_conv2d_0_group)
+ self.module5_1 = Module5(conv2d_0_in_channels=module5_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_1_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_1_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_1_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_1_conv2d_0_pad_mode,
+ conv2d_0_group=module5_1_conv2d_0_group)
+
+ def construct(self, x):
+ module5_0_opt = self.module5_0(x)
+ module5_1_opt = self.module5_1(module5_0_opt)
+ return module5_1_opt
+
+
+class Module0(nn.Cell):
+ def __init__(self, avgpool2d_0_kernel_size, conv2d_1_in_channels, conv2d_1_out_channels, conv2d_3_in_channels,
+ conv2d_3_out_channels, reshape_4_shape):
+ super(Module0, self).__init__()
+ self.avgpool2d_0 = nn.AvgPool2d(kernel_size=avgpool2d_0_kernel_size)
+ self.conv2d_1 = nn.Conv2d(in_channels=conv2d_1_in_channels,
+ out_channels=conv2d_1_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=conv2d_3_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.reshape_4 = P.Reshape()
+ self.reshape_4_shape = tuple(reshape_4_shape)
+ self.transpose_5 = P.Transpose()
+
+ def construct(self, x):
+ opt_avgpool2d_0 = self.avgpool2d_0(x)
+ opt_conv2d_1 = self.conv2d_1(opt_avgpool2d_0)
+ opt_relu_2 = self.relu_2(opt_conv2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_reshape_4 = self.reshape_4(opt_conv2d_3, self.reshape_4_shape)
+ opt_transpose_5 = self.transpose_5(opt_reshape_4, (0, 3, 1, 2))
+ return opt_transpose_5
+
+
+class Module20(nn.Cell):
+ def __init__(self, reshape_2_shape, module0_0_avgpool2d_0_kernel_size, module0_0_conv2d_1_in_channels,
+ module0_0_conv2d_1_out_channels, module0_0_conv2d_3_in_channels, module0_0_conv2d_3_out_channels,
+ module0_0_reshape_4_shape):
+ super(Module20, self).__init__()
+ self.module0_0 = Module0(avgpool2d_0_kernel_size=module0_0_avgpool2d_0_kernel_size,
+ conv2d_1_in_channels=module0_0_conv2d_1_in_channels,
+ conv2d_1_out_channels=module0_0_conv2d_1_out_channels,
+ conv2d_3_in_channels=module0_0_conv2d_3_in_channels,
+ conv2d_3_out_channels=module0_0_conv2d_3_out_channels,
+ reshape_4_shape=module0_0_reshape_4_shape)
+ self.softmax_0 = nn.Softmax(axis=3)
+ self.transpose_1 = P.Transpose()
+ self.reshape_2 = P.Reshape()
+ self.reshape_2_shape = tuple(reshape_2_shape)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ opt_softmax_0 = self.softmax_0(module0_0_opt)
+ opt_transpose_1 = self.transpose_1(opt_softmax_0, (0, 3, 2, 1))
+ opt_reshape_2 = self.reshape_2(opt_transpose_1, self.reshape_2_shape)
+ return opt_reshape_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(3, 3),
+ stride=(2, 2),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.module22_0 = Module22(module5_0_conv2d_0_in_channels=64,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(3, 3),
+ module5_0_conv2d_0_padding=(1, 1, 1, 1),
+ module5_0_conv2d_0_pad_mode="pad",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=1)
+ self.pad_maxpool2d_6 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_6 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module22_1 = Module22(module5_0_conv2d_0_in_channels=128,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_13 = P.Split(axis=1, output_num=2)
+ self.add_14_bias = 0.0
+ self.module20_0 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(80, 80),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_25 = P.Reshape()
+ self.reshape_25_shape = tuple([1, 128, 1, 1])
+ self.split_26 = P.Split(axis=1, output_num=2)
+ self.add_29_bias = 0.0
+ self.conv2d_31 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_8 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_8 = nn.AvgPool2d(kernel_size=(1, 1), stride=(1, 1))
+ self.conv2d_10 = nn.Conv2d(in_channels=128,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_33 = nn.ReLU()
+ self.module22_2 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_38 = P.Split(axis=1, output_num=2)
+ self.add_39_bias = 0.0
+ self.module20_1 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(80, 80),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_50 = P.Reshape()
+ self.reshape_50_shape = tuple([1, 128, 1, 1])
+ self.split_51 = P.Split(axis=1, output_num=2)
+ self.add_54_bias = 0.0
+ self.conv2d_56 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_58 = nn.ReLU()
+ self.module22_3 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_63 = P.Split(axis=1, output_num=2)
+ self.add_64_bias = 0.0
+ self.module20_2 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(80, 80),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_75 = P.Reshape()
+ self.reshape_75_shape = tuple([1, 128, 1, 1])
+ self.split_76 = P.Split(axis=1, output_num=2)
+ self.add_79_bias = 0.0
+ self.conv2d_81 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_83 = nn.ReLU()
+ self.module22_4 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_90 = P.Split(axis=1, output_num=2)
+ self.add_91_bias = 0.0
+ self.module20_3 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(80, 80),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_102 = P.Reshape()
+ self.reshape_102_shape = tuple([1, 256, 1, 1])
+ self.split_103 = P.Split(axis=1, output_num=2)
+ self.add_106_bias = 0.0
+ self.pad_108 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_109 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_109 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_110 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_85 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_85 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_87 = nn.Conv2d(in_channels=256,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_112 = nn.ReLU()
+ self.module22_5 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_117 = P.Split(axis=1, output_num=2)
+ self.add_118_bias = 0.0
+ self.module20_4 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_129 = P.Reshape()
+ self.reshape_129_shape = tuple([1, 256, 1, 1])
+ self.split_130 = P.Split(axis=1, output_num=2)
+ self.add_133_bias = 0.0
+ self.conv2d_135 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_137 = nn.ReLU()
+ self.module22_6 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_142 = P.Split(axis=1, output_num=2)
+ self.add_143_bias = 0.0
+ self.module20_5 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_154 = P.Reshape()
+ self.reshape_154_shape = tuple([1, 256, 1, 1])
+ self.split_155 = P.Split(axis=1, output_num=2)
+ self.add_158_bias = 0.0
+ self.conv2d_160 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_162 = nn.ReLU()
+ self.module22_7 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_167 = P.Split(axis=1, output_num=2)
+ self.add_168_bias = 0.0
+ self.module20_6 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_179 = P.Reshape()
+ self.reshape_179_shape = tuple([1, 256, 1, 1])
+ self.split_180 = P.Split(axis=1, output_num=2)
+ self.add_183_bias = 0.0
+ self.conv2d_185 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_187 = nn.ReLU()
+ self.module22_8 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_192 = P.Split(axis=1, output_num=2)
+ self.add_193_bias = 0.0
+ self.module20_7 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_204 = P.Reshape()
+ self.reshape_204_shape = tuple([1, 256, 1, 1])
+ self.split_205 = P.Split(axis=1, output_num=2)
+ self.add_208_bias = 0.0
+ self.conv2d_210 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_212 = nn.ReLU()
+ self.module22_9 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_217 = P.Split(axis=1, output_num=2)
+ self.add_218_bias = 0.0
+ self.module20_8 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_229 = P.Reshape()
+ self.reshape_229_shape = tuple([1, 256, 1, 1])
+ self.split_230 = P.Split(axis=1, output_num=2)
+ self.add_233_bias = 0.0
+ self.conv2d_235 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_237 = nn.ReLU()
+ self.module22_10 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_242 = P.Split(axis=1, output_num=2)
+ self.add_243_bias = 0.0
+ self.module20_9 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_254 = P.Reshape()
+ self.reshape_254_shape = tuple([1, 256, 1, 1])
+ self.split_255 = P.Split(axis=1, output_num=2)
+ self.add_258_bias = 0.0
+ self.conv2d_260 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_262 = nn.ReLU()
+ self.module22_11 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_267 = P.Split(axis=1, output_num=2)
+ self.add_268_bias = 0.0
+ self.module20_10 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_279 = P.Reshape()
+ self.reshape_279_shape = tuple([1, 256, 1, 1])
+ self.split_280 = P.Split(axis=1, output_num=2)
+ self.add_283_bias = 0.0
+ self.conv2d_285 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_287 = nn.ReLU()
+ self.module22_12 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_292 = P.Split(axis=1, output_num=2)
+ self.add_293_bias = 0.0
+ self.module20_11 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_304 = P.Reshape()
+ self.reshape_304_shape = tuple([1, 256, 1, 1])
+ self.split_305 = P.Split(axis=1, output_num=2)
+ self.add_308_bias = 0.0
+ self.conv2d_310 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_312 = nn.ReLU()
+ self.module22_13 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_317 = P.Split(axis=1, output_num=2)
+ self.add_318_bias = 0.0
+ self.module20_12 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_329 = P.Reshape()
+ self.reshape_329_shape = tuple([1, 256, 1, 1])
+ self.split_330 = P.Split(axis=1, output_num=2)
+ self.add_333_bias = 0.0
+ self.conv2d_335 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_337 = nn.ReLU()
+ self.module22_14 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_342 = P.Split(axis=1, output_num=2)
+ self.add_343_bias = 0.0
+ self.module20_13 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_354 = P.Reshape()
+ self.reshape_354_shape = tuple([1, 256, 1, 1])
+ self.split_355 = P.Split(axis=1, output_num=2)
+ self.add_358_bias = 0.0
+ self.conv2d_360 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_362 = nn.ReLU()
+ self.module22_15 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_367 = P.Split(axis=1, output_num=2)
+ self.add_368_bias = 0.0
+ self.module20_14 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_379 = P.Reshape()
+ self.reshape_379_shape = tuple([1, 256, 1, 1])
+ self.split_380 = P.Split(axis=1, output_num=2)
+ self.add_383_bias = 0.0
+ self.conv2d_385 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_387 = nn.ReLU()
+ self.module22_16 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_392 = P.Split(axis=1, output_num=2)
+ self.add_393_bias = 0.0
+ self.module20_15 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_404 = P.Reshape()
+ self.reshape_404_shape = tuple([1, 256, 1, 1])
+ self.split_405 = P.Split(axis=1, output_num=2)
+ self.add_408_bias = 0.0
+ self.conv2d_410 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_412 = nn.ReLU()
+ self.module22_17 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_417 = P.Split(axis=1, output_num=2)
+ self.add_418_bias = 0.0
+ self.module20_16 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_429 = P.Reshape()
+ self.reshape_429_shape = tuple([1, 256, 1, 1])
+ self.split_430 = P.Split(axis=1, output_num=2)
+ self.add_433_bias = 0.0
+ self.conv2d_435 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_437 = nn.ReLU()
+ self.module22_18 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_442 = P.Split(axis=1, output_num=2)
+ self.add_443_bias = 0.0
+ self.module20_17 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_454 = P.Reshape()
+ self.reshape_454_shape = tuple([1, 256, 1, 1])
+ self.split_455 = P.Split(axis=1, output_num=2)
+ self.add_458_bias = 0.0
+ self.conv2d_460 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_462 = nn.ReLU()
+ self.module22_19 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_467 = P.Split(axis=1, output_num=2)
+ self.add_468_bias = 0.0
+ self.module20_18 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_479 = P.Reshape()
+ self.reshape_479_shape = tuple([1, 256, 1, 1])
+ self.split_480 = P.Split(axis=1, output_num=2)
+ self.add_483_bias = 0.0
+ self.conv2d_485 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_487 = nn.ReLU()
+ self.module22_20 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_492 = P.Split(axis=1, output_num=2)
+ self.add_493_bias = 0.0
+ self.module20_19 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_504 = P.Reshape()
+ self.reshape_504_shape = tuple([1, 256, 1, 1])
+ self.split_505 = P.Split(axis=1, output_num=2)
+ self.add_508_bias = 0.0
+ self.conv2d_510 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_512 = nn.ReLU()
+ self.module22_21 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_517 = P.Split(axis=1, output_num=2)
+ self.add_518_bias = 0.0
+ self.module20_20 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_529 = P.Reshape()
+ self.reshape_529_shape = tuple([1, 256, 1, 1])
+ self.split_530 = P.Split(axis=1, output_num=2)
+ self.add_533_bias = 0.0
+ self.conv2d_535 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_537 = nn.ReLU()
+ self.module22_22 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_542 = P.Split(axis=1, output_num=2)
+ self.add_543_bias = 0.0
+ self.module20_21 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_554 = P.Reshape()
+ self.reshape_554_shape = tuple([1, 256, 1, 1])
+ self.split_555 = P.Split(axis=1, output_num=2)
+ self.add_558_bias = 0.0
+ self.conv2d_560 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_562 = nn.ReLU()
+ self.module22_23 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_567 = P.Split(axis=1, output_num=2)
+ self.add_568_bias = 0.0
+ self.module20_22 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_579 = P.Reshape()
+ self.reshape_579_shape = tuple([1, 256, 1, 1])
+ self.split_580 = P.Split(axis=1, output_num=2)
+ self.add_583_bias = 0.0
+ self.conv2d_585 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_587 = nn.ReLU()
+ self.module22_24 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_592 = P.Split(axis=1, output_num=2)
+ self.add_593_bias = 0.0
+ self.module20_23 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_604 = P.Reshape()
+ self.reshape_604_shape = tuple([1, 256, 1, 1])
+ self.split_605 = P.Split(axis=1, output_num=2)
+ self.add_608_bias = 0.0
+ self.conv2d_610 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_612 = nn.ReLU()
+ self.module22_25 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_617 = P.Split(axis=1, output_num=2)
+ self.add_618_bias = 0.0
+ self.module20_24 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_629 = P.Reshape()
+ self.reshape_629_shape = tuple([1, 256, 1, 1])
+ self.split_630 = P.Split(axis=1, output_num=2)
+ self.add_633_bias = 0.0
+ self.conv2d_635 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_637 = nn.ReLU()
+ self.module22_26 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_642 = P.Split(axis=1, output_num=2)
+ self.add_643_bias = 0.0
+ self.module20_25 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_654 = P.Reshape()
+ self.reshape_654_shape = tuple([1, 256, 1, 1])
+ self.split_655 = P.Split(axis=1, output_num=2)
+ self.add_658_bias = 0.0
+ self.conv2d_660 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_662 = nn.ReLU()
+ self.module22_27 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_667 = P.Split(axis=1, output_num=2)
+ self.add_668_bias = 0.0
+ self.module20_26 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_679 = P.Reshape()
+ self.reshape_679_shape = tuple([1, 256, 1, 1])
+ self.split_680 = P.Split(axis=1, output_num=2)
+ self.add_683_bias = 0.0
+ self.conv2d_685 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_687 = nn.ReLU()
+ self.module22_28 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_694 = P.Split(axis=1, output_num=2)
+ self.add_695_bias = 0.0
+ self.module20_27 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(40, 40),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_706 = P.Reshape()
+ self.reshape_706_shape = tuple([1, 512, 1, 1])
+ self.split_707 = P.Split(axis=1, output_num=2)
+ self.add_710_bias = 0.0
+ self.pad_712 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_713 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_713 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_714 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_689 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_689 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_691 = nn.Conv2d(in_channels=512,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_716 = nn.ReLU()
+ self.module22_29 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_721 = P.Split(axis=1, output_num=2)
+ self.add_722_bias = 0.0
+ self.module20_28 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_733 = P.Reshape()
+ self.reshape_733_shape = tuple([1, 512, 1, 1])
+ self.split_734 = P.Split(axis=1, output_num=2)
+ self.add_737_bias = 0.0
+ self.conv2d_739 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_741 = nn.ReLU()
+ self.module22_30 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_746 = P.Split(axis=1, output_num=2)
+ self.add_747_bias = 0.0
+ self.module20_29 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_758 = P.Reshape()
+ self.reshape_758_shape = tuple([1, 512, 1, 1])
+ self.split_759 = P.Split(axis=1, output_num=2)
+ self.add_762_bias = 0.0
+ self.conv2d_764 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_766 = nn.ReLU()
+ self.module22_31 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_771 = P.Split(axis=1, output_num=2)
+ self.add_772_bias = 0.0
+ self.module20_30 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_783 = P.Reshape()
+ self.reshape_783_shape = tuple([1, 512, 1, 1])
+ self.split_784 = P.Split(axis=1, output_num=2)
+ self.add_787_bias = 0.0
+ self.conv2d_789 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_791 = nn.ReLU()
+ self.module22_32 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_796 = P.Split(axis=1, output_num=2)
+ self.add_797_bias = 0.0
+ self.module20_31 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_808 = P.Reshape()
+ self.reshape_808_shape = tuple([1, 512, 1, 1])
+ self.split_809 = P.Split(axis=1, output_num=2)
+ self.add_812_bias = 0.0
+ self.conv2d_814 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_816 = nn.ReLU()
+ self.module22_33 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_821 = P.Split(axis=1, output_num=2)
+ self.add_822_bias = 0.0
+ self.module20_32 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_833 = P.Reshape()
+ self.reshape_833_shape = tuple([1, 512, 1, 1])
+ self.split_834 = P.Split(axis=1, output_num=2)
+ self.add_837_bias = 0.0
+ self.conv2d_839 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_841 = nn.ReLU()
+ self.module22_34 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_846 = P.Split(axis=1, output_num=2)
+ self.add_847_bias = 0.0
+ self.module20_33 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_858 = P.Reshape()
+ self.reshape_858_shape = tuple([1, 512, 1, 1])
+ self.split_859 = P.Split(axis=1, output_num=2)
+ self.add_862_bias = 0.0
+ self.conv2d_864 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_866 = nn.ReLU()
+ self.module22_35 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_871 = P.Split(axis=1, output_num=2)
+ self.add_872_bias = 0.0
+ self.module20_34 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_883 = P.Reshape()
+ self.reshape_883_shape = tuple([1, 512, 1, 1])
+ self.split_884 = P.Split(axis=1, output_num=2)
+ self.add_887_bias = 0.0
+ self.conv2d_889 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_891 = nn.ReLU()
+ self.module22_36 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_896 = P.Split(axis=1, output_num=2)
+ self.add_897_bias = 0.0
+ self.module20_35 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_908 = P.Reshape()
+ self.reshape_908_shape = tuple([1, 512, 1, 1])
+ self.split_909 = P.Split(axis=1, output_num=2)
+ self.add_912_bias = 0.0
+ self.conv2d_914 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_916 = nn.ReLU()
+ self.module22_37 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_921 = P.Split(axis=1, output_num=2)
+ self.add_922_bias = 0.0
+ self.module20_36 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_933 = P.Reshape()
+ self.reshape_933_shape = tuple([1, 512, 1, 1])
+ self.split_934 = P.Split(axis=1, output_num=2)
+ self.add_937_bias = 0.0
+ self.conv2d_939 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_941 = nn.ReLU()
+ self.module22_38 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_946 = P.Split(axis=1, output_num=2)
+ self.add_947_bias = 0.0
+ self.module20_37 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_958 = P.Reshape()
+ self.reshape_958_shape = tuple([1, 512, 1, 1])
+ self.split_959 = P.Split(axis=1, output_num=2)
+ self.add_962_bias = 0.0
+ self.conv2d_964 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_966 = nn.ReLU()
+ self.module22_39 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_971 = P.Split(axis=1, output_num=2)
+ self.add_972_bias = 0.0
+ self.module20_38 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_983 = P.Reshape()
+ self.reshape_983_shape = tuple([1, 512, 1, 1])
+ self.split_984 = P.Split(axis=1, output_num=2)
+ self.add_987_bias = 0.0
+ self.conv2d_989 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_991 = nn.ReLU()
+ self.module22_40 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_996 = P.Split(axis=1, output_num=2)
+ self.add_997_bias = 0.0
+ self.module20_39 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1008 = P.Reshape()
+ self.reshape_1008_shape = tuple([1, 512, 1, 1])
+ self.split_1009 = P.Split(axis=1, output_num=2)
+ self.add_1012_bias = 0.0
+ self.conv2d_1014 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1016 = nn.ReLU()
+ self.module22_41 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1021 = P.Split(axis=1, output_num=2)
+ self.add_1022_bias = 0.0
+ self.module20_40 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1033 = P.Reshape()
+ self.reshape_1033_shape = tuple([1, 512, 1, 1])
+ self.split_1034 = P.Split(axis=1, output_num=2)
+ self.add_1037_bias = 0.0
+ self.conv2d_1039 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1041 = nn.ReLU()
+ self.module22_42 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1046 = P.Split(axis=1, output_num=2)
+ self.add_1047_bias = 0.0
+ self.module20_41 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1058 = P.Reshape()
+ self.reshape_1058_shape = tuple([1, 512, 1, 1])
+ self.split_1059 = P.Split(axis=1, output_num=2)
+ self.add_1062_bias = 0.0
+ self.conv2d_1064 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1066 = nn.ReLU()
+ self.module22_43 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1071 = P.Split(axis=1, output_num=2)
+ self.add_1072_bias = 0.0
+ self.module20_42 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1083 = P.Reshape()
+ self.reshape_1083_shape = tuple([1, 512, 1, 1])
+ self.split_1084 = P.Split(axis=1, output_num=2)
+ self.add_1087_bias = 0.0
+ self.conv2d_1089 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1091 = nn.ReLU()
+ self.module22_44 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1096 = P.Split(axis=1, output_num=2)
+ self.add_1097_bias = 0.0
+ self.module20_43 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1108 = P.Reshape()
+ self.reshape_1108_shape = tuple([1, 512, 1, 1])
+ self.split_1109 = P.Split(axis=1, output_num=2)
+ self.add_1112_bias = 0.0
+ self.conv2d_1114 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1116 = nn.ReLU()
+ self.module22_45 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1121 = P.Split(axis=1, output_num=2)
+ self.add_1122_bias = 0.0
+ self.module20_44 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1133 = P.Reshape()
+ self.reshape_1133_shape = tuple([1, 512, 1, 1])
+ self.split_1134 = P.Split(axis=1, output_num=2)
+ self.add_1137_bias = 0.0
+ self.conv2d_1139 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1141 = nn.ReLU()
+ self.module22_46 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1146 = P.Split(axis=1, output_num=2)
+ self.add_1147_bias = 0.0
+ self.module20_45 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1158 = P.Reshape()
+ self.reshape_1158_shape = tuple([1, 512, 1, 1])
+ self.split_1159 = P.Split(axis=1, output_num=2)
+ self.add_1162_bias = 0.0
+ self.conv2d_1164 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1166 = nn.ReLU()
+ self.module22_47 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1171 = P.Split(axis=1, output_num=2)
+ self.add_1172_bias = 0.0
+ self.module20_46 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1183 = P.Reshape()
+ self.reshape_1183_shape = tuple([1, 512, 1, 1])
+ self.split_1184 = P.Split(axis=1, output_num=2)
+ self.add_1187_bias = 0.0
+ self.conv2d_1189 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1191 = nn.ReLU()
+ self.module22_48 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1196 = P.Split(axis=1, output_num=2)
+ self.add_1197_bias = 0.0
+ self.module20_47 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1208 = P.Reshape()
+ self.reshape_1208_shape = tuple([1, 512, 1, 1])
+ self.split_1209 = P.Split(axis=1, output_num=2)
+ self.add_1212_bias = 0.0
+ self.conv2d_1214 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1216 = nn.ReLU()
+ self.module22_49 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1221 = P.Split(axis=1, output_num=2)
+ self.add_1222_bias = 0.0
+ self.module20_48 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1233 = P.Reshape()
+ self.reshape_1233_shape = tuple([1, 512, 1, 1])
+ self.split_1234 = P.Split(axis=1, output_num=2)
+ self.add_1237_bias = 0.0
+ self.conv2d_1239 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1241 = nn.ReLU()
+ self.module22_50 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1246 = P.Split(axis=1, output_num=2)
+ self.add_1247_bias = 0.0
+ self.module20_49 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1258 = P.Reshape()
+ self.reshape_1258_shape = tuple([1, 512, 1, 1])
+ self.split_1259 = P.Split(axis=1, output_num=2)
+ self.add_1262_bias = 0.0
+ self.conv2d_1264 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1266 = nn.ReLU()
+ self.module22_51 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1271 = P.Split(axis=1, output_num=2)
+ self.add_1272_bias = 0.0
+ self.module20_50 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1283 = P.Reshape()
+ self.reshape_1283_shape = tuple([1, 512, 1, 1])
+ self.split_1284 = P.Split(axis=1, output_num=2)
+ self.add_1287_bias = 0.0
+ self.conv2d_1289 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1291 = nn.ReLU()
+ self.module22_52 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1296 = P.Split(axis=1, output_num=2)
+ self.add_1297_bias = 0.0
+ self.module20_51 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1308 = P.Reshape()
+ self.reshape_1308_shape = tuple([1, 512, 1, 1])
+ self.split_1309 = P.Split(axis=1, output_num=2)
+ self.add_1312_bias = 0.0
+ self.conv2d_1314 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1316 = nn.ReLU()
+ self.module22_53 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1321 = P.Split(axis=1, output_num=2)
+ self.add_1322_bias = 0.0
+ self.module20_52 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1333 = P.Reshape()
+ self.reshape_1333_shape = tuple([1, 512, 1, 1])
+ self.split_1334 = P.Split(axis=1, output_num=2)
+ self.add_1337_bias = 0.0
+ self.conv2d_1339 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1341 = nn.ReLU()
+ self.module22_54 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1346 = P.Split(axis=1, output_num=2)
+ self.add_1347_bias = 0.0
+ self.module20_53 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1358 = P.Reshape()
+ self.reshape_1358_shape = tuple([1, 512, 1, 1])
+ self.split_1359 = P.Split(axis=1, output_num=2)
+ self.add_1362_bias = 0.0
+ self.conv2d_1364 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1366 = nn.ReLU()
+ self.module22_55 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1371 = P.Split(axis=1, output_num=2)
+ self.add_1372_bias = 0.0
+ self.module20_54 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1383 = P.Reshape()
+ self.reshape_1383_shape = tuple([1, 512, 1, 1])
+ self.split_1384 = P.Split(axis=1, output_num=2)
+ self.add_1387_bias = 0.0
+ self.conv2d_1389 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1391 = nn.ReLU()
+ self.module22_56 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1396 = P.Split(axis=1, output_num=2)
+ self.add_1397_bias = 0.0
+ self.module20_55 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1408 = P.Reshape()
+ self.reshape_1408_shape = tuple([1, 512, 1, 1])
+ self.split_1409 = P.Split(axis=1, output_num=2)
+ self.add_1412_bias = 0.0
+ self.conv2d_1414 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1416 = nn.ReLU()
+ self.module22_57 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1421 = P.Split(axis=1, output_num=2)
+ self.add_1422_bias = 0.0
+ self.module20_56 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1433 = P.Reshape()
+ self.reshape_1433_shape = tuple([1, 512, 1, 1])
+ self.split_1434 = P.Split(axis=1, output_num=2)
+ self.add_1437_bias = 0.0
+ self.conv2d_1439 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1441 = nn.ReLU()
+ self.module22_58 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1446 = P.Split(axis=1, output_num=2)
+ self.add_1447_bias = 0.0
+ self.module20_57 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1458 = P.Reshape()
+ self.reshape_1458_shape = tuple([1, 512, 1, 1])
+ self.split_1459 = P.Split(axis=1, output_num=2)
+ self.add_1462_bias = 0.0
+ self.conv2d_1464 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1466 = nn.ReLU()
+ self.module22_59 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1471 = P.Split(axis=1, output_num=2)
+ self.add_1472_bias = 0.0
+ self.module20_58 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1483 = P.Reshape()
+ self.reshape_1483_shape = tuple([1, 512, 1, 1])
+ self.split_1484 = P.Split(axis=1, output_num=2)
+ self.add_1487_bias = 0.0
+ self.conv2d_1489 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1491 = nn.ReLU()
+ self.module22_60 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1496 = P.Split(axis=1, output_num=2)
+ self.add_1497_bias = 0.0
+ self.module20_59 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1508 = P.Reshape()
+ self.reshape_1508_shape = tuple([1, 512, 1, 1])
+ self.split_1509 = P.Split(axis=1, output_num=2)
+ self.add_1512_bias = 0.0
+ self.conv2d_1514 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1516 = nn.ReLU()
+ self.module22_61 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1521 = P.Split(axis=1, output_num=2)
+ self.add_1522_bias = 0.0
+ self.module20_60 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1533 = P.Reshape()
+ self.reshape_1533_shape = tuple([1, 512, 1, 1])
+ self.split_1534 = P.Split(axis=1, output_num=2)
+ self.add_1537_bias = 0.0
+ self.conv2d_1539 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1541 = nn.ReLU()
+ self.module22_62 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1546 = P.Split(axis=1, output_num=2)
+ self.add_1547_bias = 0.0
+ self.module20_61 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1558 = P.Reshape()
+ self.reshape_1558_shape = tuple([1, 512, 1, 1])
+ self.split_1559 = P.Split(axis=1, output_num=2)
+ self.add_1562_bias = 0.0
+ self.conv2d_1564 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1566 = nn.ReLU()
+ self.module22_63 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1571 = P.Split(axis=1, output_num=2)
+ self.add_1572_bias = 0.0
+ self.module20_62 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1583 = P.Reshape()
+ self.reshape_1583_shape = tuple([1, 512, 1, 1])
+ self.split_1584 = P.Split(axis=1, output_num=2)
+ self.add_1587_bias = 0.0
+ self.conv2d_1589 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1591 = nn.ReLU()
+ self.module22_64 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1598 = P.Split(axis=1, output_num=2)
+ self.add_1599_bias = 0.0
+ self.module20_63 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(20, 20),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_1610 = P.Reshape()
+ self.reshape_1610_shape = tuple([1, 1024, 1, 1])
+ self.split_1611 = P.Split(axis=1, output_num=2)
+ self.add_1614_bias = 0.0
+ self.pad_1616 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_1617 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_1617 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_1618 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_1593 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_1593 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_1595 = nn.Conv2d(in_channels=1024,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1620 = nn.ReLU()
+ self.module22_65 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1625 = P.Split(axis=1, output_num=2)
+ self.add_1626_bias = 0.0
+ self.module20_64 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(10, 10),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_1637 = P.Reshape()
+ self.reshape_1637_shape = tuple([1, 1024, 1, 1])
+ self.split_1638 = P.Split(axis=1, output_num=2)
+ self.add_1641_bias = 0.0
+ self.conv2d_1643 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1645 = nn.ReLU()
+ self.module22_66 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1650 = P.Split(axis=1, output_num=2)
+ self.add_1651_bias = 0.0
+ self.module20_65 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(10, 10),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_1662 = P.Reshape()
+ self.reshape_1662_shape = tuple([1, 1024, 1, 1])
+ self.split_1663 = P.Split(axis=1, output_num=2)
+ self.add_1666_bias = 0.0
+ self.conv2d_1668 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1670 = nn.ReLU()
+ self.avgpool2d_1671 = nn.AvgPool2d(kernel_size=(10, 10))
+ self.reshape_1672 = P.Reshape()
+ self.reshape_1672_shape = tuple([1, 2048])
+ self.flatten_1673 = nn.Flatten()
+ self.dense_1674 = nn.Dense(in_channels=2048, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ module22_0_opt = self.module22_0(opt_relu_1)
+ opt_maxpool2d_6 = self.pad_maxpool2d_6(module22_0_opt)
+ opt_maxpool2d_6 = self.maxpool2d_6(opt_maxpool2d_6)
+ module22_1_opt = self.module22_1(opt_maxpool2d_6)
+ opt_split_13, opt_split_13_1 = self.split_13(module22_1_opt)
+ opt_add_14 = opt_split_13 + self.add_14_bias
+ opt_add_15 = P.Add()(opt_add_14, opt_split_13_1)
+ module20_0_opt = self.module20_0(opt_add_15)
+ opt_reshape_25 = self.reshape_25(module20_0_opt, self.reshape_25_shape)
+ opt_split_26, opt_split_26_1 = self.split_26(opt_reshape_25)
+ opt_mul_27 = P.Mul()(opt_split_26, opt_split_13)
+ opt_mul_28 = P.Mul()(opt_split_26_1, opt_split_13_1)
+ opt_add_29 = opt_mul_27 + self.add_29_bias
+ opt_add_30 = P.Add()(opt_add_29, opt_mul_28)
+ opt_conv2d_31 = self.conv2d_31(opt_add_30)
+ opt_avgpool2d_8 = self.pad_avgpool2d_8(opt_maxpool2d_6)
+ opt_avgpool2d_8 = self.avgpool2d_8(opt_avgpool2d_8)
+ opt_conv2d_10 = self.conv2d_10(opt_avgpool2d_8)
+ opt_add_32 = P.Add()(opt_conv2d_31, opt_conv2d_10)
+ opt_relu_33 = self.relu_33(opt_add_32)
+ module22_2_opt = self.module22_2(opt_relu_33)
+ opt_split_38, opt_split_38_1 = self.split_38(module22_2_opt)
+ opt_add_39 = opt_split_38 + self.add_39_bias
+ opt_add_40 = P.Add()(opt_add_39, opt_split_38_1)
+ module20_1_opt = self.module20_1(opt_add_40)
+ opt_reshape_50 = self.reshape_50(module20_1_opt, self.reshape_50_shape)
+ opt_split_51, opt_split_51_1 = self.split_51(opt_reshape_50)
+ opt_mul_52 = P.Mul()(opt_split_51, opt_split_38)
+ opt_mul_53 = P.Mul()(opt_split_51_1, opt_split_38_1)
+ opt_add_54 = opt_mul_52 + self.add_54_bias
+ opt_add_55 = P.Add()(opt_add_54, opt_mul_53)
+ opt_conv2d_56 = self.conv2d_56(opt_add_55)
+ opt_add_57 = P.Add()(opt_conv2d_56, opt_relu_33)
+ opt_relu_58 = self.relu_58(opt_add_57)
+ module22_3_opt = self.module22_3(opt_relu_58)
+ opt_split_63, opt_split_63_1 = self.split_63(module22_3_opt)
+ opt_add_64 = opt_split_63 + self.add_64_bias
+ opt_add_65 = P.Add()(opt_add_64, opt_split_63_1)
+ module20_2_opt = self.module20_2(opt_add_65)
+ opt_reshape_75 = self.reshape_75(module20_2_opt, self.reshape_75_shape)
+ opt_split_76, opt_split_76_1 = self.split_76(opt_reshape_75)
+ opt_mul_77 = P.Mul()(opt_split_76, opt_split_63)
+ opt_mul_78 = P.Mul()(opt_split_76_1, opt_split_63_1)
+ opt_add_79 = opt_mul_77 + self.add_79_bias
+ opt_add_80 = P.Add()(opt_add_79, opt_mul_78)
+ opt_conv2d_81 = self.conv2d_81(opt_add_80)
+ opt_add_82 = P.Add()(opt_conv2d_81, opt_relu_58)
+ opt_relu_83 = self.relu_83(opt_add_82)
+ module22_4_opt = self.module22_4(opt_relu_83)
+ opt_split_90, opt_split_90_1 = self.split_90(module22_4_opt)
+ opt_add_91 = opt_split_90 + self.add_91_bias
+ opt_add_92 = P.Add()(opt_add_91, opt_split_90_1)
+ module20_3_opt = self.module20_3(opt_add_92)
+ opt_reshape_102 = self.reshape_102(module20_3_opt, self.reshape_102_shape)
+ opt_split_103, opt_split_103_1 = self.split_103(opt_reshape_102)
+ opt_mul_104 = P.Mul()(opt_split_103, opt_split_90)
+ opt_mul_105 = P.Mul()(opt_split_103_1, opt_split_90_1)
+ opt_add_106 = opt_mul_104 + self.add_106_bias
+ opt_add_107 = P.Add()(opt_add_106, opt_mul_105)
+ opt_pad_108 = self.pad_108(opt_add_107)
+ opt_avgpool2d_109 = self.pad_avgpool2d_109(opt_pad_108)
+ opt_avgpool2d_109 = self.avgpool2d_109(opt_avgpool2d_109)
+ opt_conv2d_110 = self.conv2d_110(opt_avgpool2d_109)
+ opt_avgpool2d_85 = self.pad_avgpool2d_85(opt_relu_83)
+ opt_avgpool2d_85 = self.avgpool2d_85(opt_avgpool2d_85)
+ opt_conv2d_87 = self.conv2d_87(opt_avgpool2d_85)
+ opt_add_111 = P.Add()(opt_conv2d_110, opt_conv2d_87)
+ opt_relu_112 = self.relu_112(opt_add_111)
+ module22_5_opt = self.module22_5(opt_relu_112)
+ opt_split_117, opt_split_117_1 = self.split_117(module22_5_opt)
+ opt_add_118 = opt_split_117 + self.add_118_bias
+ opt_add_119 = P.Add()(opt_add_118, opt_split_117_1)
+ module20_4_opt = self.module20_4(opt_add_119)
+ opt_reshape_129 = self.reshape_129(module20_4_opt, self.reshape_129_shape)
+ opt_split_130, opt_split_130_1 = self.split_130(opt_reshape_129)
+ opt_mul_131 = P.Mul()(opt_split_130, opt_split_117)
+ opt_mul_132 = P.Mul()(opt_split_130_1, opt_split_117_1)
+ opt_add_133 = opt_mul_131 + self.add_133_bias
+ opt_add_134 = P.Add()(opt_add_133, opt_mul_132)
+ opt_conv2d_135 = self.conv2d_135(opt_add_134)
+ opt_add_136 = P.Add()(opt_conv2d_135, opt_relu_112)
+ opt_relu_137 = self.relu_137(opt_add_136)
+ module22_6_opt = self.module22_6(opt_relu_137)
+ opt_split_142, opt_split_142_1 = self.split_142(module22_6_opt)
+ opt_add_143 = opt_split_142 + self.add_143_bias
+ opt_add_144 = P.Add()(opt_add_143, opt_split_142_1)
+ module20_5_opt = self.module20_5(opt_add_144)
+ opt_reshape_154 = self.reshape_154(module20_5_opt, self.reshape_154_shape)
+ opt_split_155, opt_split_155_1 = self.split_155(opt_reshape_154)
+ opt_mul_156 = P.Mul()(opt_split_155, opt_split_142)
+ opt_mul_157 = P.Mul()(opt_split_155_1, opt_split_142_1)
+ opt_add_158 = opt_mul_156 + self.add_158_bias
+ opt_add_159 = P.Add()(opt_add_158, opt_mul_157)
+ opt_conv2d_160 = self.conv2d_160(opt_add_159)
+ opt_add_161 = P.Add()(opt_conv2d_160, opt_relu_137)
+ opt_relu_162 = self.relu_162(opt_add_161)
+ module22_7_opt = self.module22_7(opt_relu_162)
+ opt_split_167, opt_split_167_1 = self.split_167(module22_7_opt)
+ opt_add_168 = opt_split_167 + self.add_168_bias
+ opt_add_169 = P.Add()(opt_add_168, opt_split_167_1)
+ module20_6_opt = self.module20_6(opt_add_169)
+ opt_reshape_179 = self.reshape_179(module20_6_opt, self.reshape_179_shape)
+ opt_split_180, opt_split_180_1 = self.split_180(opt_reshape_179)
+ opt_mul_181 = P.Mul()(opt_split_180, opt_split_167)
+ opt_mul_182 = P.Mul()(opt_split_180_1, opt_split_167_1)
+ opt_add_183 = opt_mul_181 + self.add_183_bias
+ opt_add_184 = P.Add()(opt_add_183, opt_mul_182)
+ opt_conv2d_185 = self.conv2d_185(opt_add_184)
+ opt_add_186 = P.Add()(opt_conv2d_185, opt_relu_162)
+ opt_relu_187 = self.relu_187(opt_add_186)
+ module22_8_opt = self.module22_8(opt_relu_187)
+ opt_split_192, opt_split_192_1 = self.split_192(module22_8_opt)
+ opt_add_193 = opt_split_192 + self.add_193_bias
+ opt_add_194 = P.Add()(opt_add_193, opt_split_192_1)
+ module20_7_opt = self.module20_7(opt_add_194)
+ opt_reshape_204 = self.reshape_204(module20_7_opt, self.reshape_204_shape)
+ opt_split_205, opt_split_205_1 = self.split_205(opt_reshape_204)
+ opt_mul_206 = P.Mul()(opt_split_205, opt_split_192)
+ opt_mul_207 = P.Mul()(opt_split_205_1, opt_split_192_1)
+ opt_add_208 = opt_mul_206 + self.add_208_bias
+ opt_add_209 = P.Add()(opt_add_208, opt_mul_207)
+ opt_conv2d_210 = self.conv2d_210(opt_add_209)
+ opt_add_211 = P.Add()(opt_conv2d_210, opt_relu_187)
+ opt_relu_212 = self.relu_212(opt_add_211)
+ module22_9_opt = self.module22_9(opt_relu_212)
+ opt_split_217, opt_split_217_1 = self.split_217(module22_9_opt)
+ opt_add_218 = opt_split_217 + self.add_218_bias
+ opt_add_219 = P.Add()(opt_add_218, opt_split_217_1)
+ module20_8_opt = self.module20_8(opt_add_219)
+ opt_reshape_229 = self.reshape_229(module20_8_opt, self.reshape_229_shape)
+ opt_split_230, opt_split_230_1 = self.split_230(opt_reshape_229)
+ opt_mul_231 = P.Mul()(opt_split_230, opt_split_217)
+ opt_mul_232 = P.Mul()(opt_split_230_1, opt_split_217_1)
+ opt_add_233 = opt_mul_231 + self.add_233_bias
+ opt_add_234 = P.Add()(opt_add_233, opt_mul_232)
+ opt_conv2d_235 = self.conv2d_235(opt_add_234)
+ opt_add_236 = P.Add()(opt_conv2d_235, opt_relu_212)
+ opt_relu_237 = self.relu_237(opt_add_236)
+ module22_10_opt = self.module22_10(opt_relu_237)
+ opt_split_242, opt_split_242_1 = self.split_242(module22_10_opt)
+ opt_add_243 = opt_split_242 + self.add_243_bias
+ opt_add_244 = P.Add()(opt_add_243, opt_split_242_1)
+ module20_9_opt = self.module20_9(opt_add_244)
+ opt_reshape_254 = self.reshape_254(module20_9_opt, self.reshape_254_shape)
+ opt_split_255, opt_split_255_1 = self.split_255(opt_reshape_254)
+ opt_mul_256 = P.Mul()(opt_split_255, opt_split_242)
+ opt_mul_257 = P.Mul()(opt_split_255_1, opt_split_242_1)
+ opt_add_258 = opt_mul_256 + self.add_258_bias
+ opt_add_259 = P.Add()(opt_add_258, opt_mul_257)
+ opt_conv2d_260 = self.conv2d_260(opt_add_259)
+ opt_add_261 = P.Add()(opt_conv2d_260, opt_relu_237)
+ opt_relu_262 = self.relu_262(opt_add_261)
+ module22_11_opt = self.module22_11(opt_relu_262)
+ opt_split_267, opt_split_267_1 = self.split_267(module22_11_opt)
+ opt_add_268 = opt_split_267 + self.add_268_bias
+ opt_add_269 = P.Add()(opt_add_268, opt_split_267_1)
+ module20_10_opt = self.module20_10(opt_add_269)
+ opt_reshape_279 = self.reshape_279(module20_10_opt, self.reshape_279_shape)
+ opt_split_280, opt_split_280_1 = self.split_280(opt_reshape_279)
+ opt_mul_281 = P.Mul()(opt_split_280, opt_split_267)
+ opt_mul_282 = P.Mul()(opt_split_280_1, opt_split_267_1)
+ opt_add_283 = opt_mul_281 + self.add_283_bias
+ opt_add_284 = P.Add()(opt_add_283, opt_mul_282)
+ opt_conv2d_285 = self.conv2d_285(opt_add_284)
+ opt_add_286 = P.Add()(opt_conv2d_285, opt_relu_262)
+ opt_relu_287 = self.relu_287(opt_add_286)
+ module22_12_opt = self.module22_12(opt_relu_287)
+ opt_split_292, opt_split_292_1 = self.split_292(module22_12_opt)
+ opt_add_293 = opt_split_292 + self.add_293_bias
+ opt_add_294 = P.Add()(opt_add_293, opt_split_292_1)
+ module20_11_opt = self.module20_11(opt_add_294)
+ opt_reshape_304 = self.reshape_304(module20_11_opt, self.reshape_304_shape)
+ opt_split_305, opt_split_305_1 = self.split_305(opt_reshape_304)
+ opt_mul_306 = P.Mul()(opt_split_305, opt_split_292)
+ opt_mul_307 = P.Mul()(opt_split_305_1, opt_split_292_1)
+ opt_add_308 = opt_mul_306 + self.add_308_bias
+ opt_add_309 = P.Add()(opt_add_308, opt_mul_307)
+ opt_conv2d_310 = self.conv2d_310(opt_add_309)
+ opt_add_311 = P.Add()(opt_conv2d_310, opt_relu_287)
+ opt_relu_312 = self.relu_312(opt_add_311)
+ module22_13_opt = self.module22_13(opt_relu_312)
+ opt_split_317, opt_split_317_1 = self.split_317(module22_13_opt)
+ opt_add_318 = opt_split_317 + self.add_318_bias
+ opt_add_319 = P.Add()(opt_add_318, opt_split_317_1)
+ module20_12_opt = self.module20_12(opt_add_319)
+ opt_reshape_329 = self.reshape_329(module20_12_opt, self.reshape_329_shape)
+ opt_split_330, opt_split_330_1 = self.split_330(opt_reshape_329)
+ opt_mul_331 = P.Mul()(opt_split_330, opt_split_317)
+ opt_mul_332 = P.Mul()(opt_split_330_1, opt_split_317_1)
+ opt_add_333 = opt_mul_331 + self.add_333_bias
+ opt_add_334 = P.Add()(opt_add_333, opt_mul_332)
+ opt_conv2d_335 = self.conv2d_335(opt_add_334)
+ opt_add_336 = P.Add()(opt_conv2d_335, opt_relu_312)
+ opt_relu_337 = self.relu_337(opt_add_336)
+ module22_14_opt = self.module22_14(opt_relu_337)
+ opt_split_342, opt_split_342_1 = self.split_342(module22_14_opt)
+ opt_add_343 = opt_split_342 + self.add_343_bias
+ opt_add_344 = P.Add()(opt_add_343, opt_split_342_1)
+ module20_13_opt = self.module20_13(opt_add_344)
+ opt_reshape_354 = self.reshape_354(module20_13_opt, self.reshape_354_shape)
+ opt_split_355, opt_split_355_1 = self.split_355(opt_reshape_354)
+ opt_mul_356 = P.Mul()(opt_split_355, opt_split_342)
+ opt_mul_357 = P.Mul()(opt_split_355_1, opt_split_342_1)
+ opt_add_358 = opt_mul_356 + self.add_358_bias
+ opt_add_359 = P.Add()(opt_add_358, opt_mul_357)
+ opt_conv2d_360 = self.conv2d_360(opt_add_359)
+ opt_add_361 = P.Add()(opt_conv2d_360, opt_relu_337)
+ opt_relu_362 = self.relu_362(opt_add_361)
+ module22_15_opt = self.module22_15(opt_relu_362)
+ opt_split_367, opt_split_367_1 = self.split_367(module22_15_opt)
+ opt_add_368 = opt_split_367 + self.add_368_bias
+ opt_add_369 = P.Add()(opt_add_368, opt_split_367_1)
+ module20_14_opt = self.module20_14(opt_add_369)
+ opt_reshape_379 = self.reshape_379(module20_14_opt, self.reshape_379_shape)
+ opt_split_380, opt_split_380_1 = self.split_380(opt_reshape_379)
+ opt_mul_381 = P.Mul()(opt_split_380, opt_split_367)
+ opt_mul_382 = P.Mul()(opt_split_380_1, opt_split_367_1)
+ opt_add_383 = opt_mul_381 + self.add_383_bias
+ opt_add_384 = P.Add()(opt_add_383, opt_mul_382)
+ opt_conv2d_385 = self.conv2d_385(opt_add_384)
+ opt_add_386 = P.Add()(opt_conv2d_385, opt_relu_362)
+ opt_relu_387 = self.relu_387(opt_add_386)
+ module22_16_opt = self.module22_16(opt_relu_387)
+ opt_split_392, opt_split_392_1 = self.split_392(module22_16_opt)
+ opt_add_393 = opt_split_392 + self.add_393_bias
+ opt_add_394 = P.Add()(opt_add_393, opt_split_392_1)
+ module20_15_opt = self.module20_15(opt_add_394)
+ opt_reshape_404 = self.reshape_404(module20_15_opt, self.reshape_404_shape)
+ opt_split_405, opt_split_405_1 = self.split_405(opt_reshape_404)
+ opt_mul_406 = P.Mul()(opt_split_405, opt_split_392)
+ opt_mul_407 = P.Mul()(opt_split_405_1, opt_split_392_1)
+ opt_add_408 = opt_mul_406 + self.add_408_bias
+ opt_add_409 = P.Add()(opt_add_408, opt_mul_407)
+ opt_conv2d_410 = self.conv2d_410(opt_add_409)
+ opt_add_411 = P.Add()(opt_conv2d_410, opt_relu_387)
+ opt_relu_412 = self.relu_412(opt_add_411)
+ module22_17_opt = self.module22_17(opt_relu_412)
+ opt_split_417, opt_split_417_1 = self.split_417(module22_17_opt)
+ opt_add_418 = opt_split_417 + self.add_418_bias
+ opt_add_419 = P.Add()(opt_add_418, opt_split_417_1)
+ module20_16_opt = self.module20_16(opt_add_419)
+ opt_reshape_429 = self.reshape_429(module20_16_opt, self.reshape_429_shape)
+ opt_split_430, opt_split_430_1 = self.split_430(opt_reshape_429)
+ opt_mul_431 = P.Mul()(opt_split_430, opt_split_417)
+ opt_mul_432 = P.Mul()(opt_split_430_1, opt_split_417_1)
+ opt_add_433 = opt_mul_431 + self.add_433_bias
+ opt_add_434 = P.Add()(opt_add_433, opt_mul_432)
+ opt_conv2d_435 = self.conv2d_435(opt_add_434)
+ opt_add_436 = P.Add()(opt_conv2d_435, opt_relu_412)
+ opt_relu_437 = self.relu_437(opt_add_436)
+ module22_18_opt = self.module22_18(opt_relu_437)
+ opt_split_442, opt_split_442_1 = self.split_442(module22_18_opt)
+ opt_add_443 = opt_split_442 + self.add_443_bias
+ opt_add_444 = P.Add()(opt_add_443, opt_split_442_1)
+ module20_17_opt = self.module20_17(opt_add_444)
+ opt_reshape_454 = self.reshape_454(module20_17_opt, self.reshape_454_shape)
+ opt_split_455, opt_split_455_1 = self.split_455(opt_reshape_454)
+ opt_mul_456 = P.Mul()(opt_split_455, opt_split_442)
+ opt_mul_457 = P.Mul()(opt_split_455_1, opt_split_442_1)
+ opt_add_458 = opt_mul_456 + self.add_458_bias
+ opt_add_459 = P.Add()(opt_add_458, opt_mul_457)
+ opt_conv2d_460 = self.conv2d_460(opt_add_459)
+ opt_add_461 = P.Add()(opt_conv2d_460, opt_relu_437)
+ opt_relu_462 = self.relu_462(opt_add_461)
+ module22_19_opt = self.module22_19(opt_relu_462)
+ opt_split_467, opt_split_467_1 = self.split_467(module22_19_opt)
+ opt_add_468 = opt_split_467 + self.add_468_bias
+ opt_add_469 = P.Add()(opt_add_468, opt_split_467_1)
+ module20_18_opt = self.module20_18(opt_add_469)
+ opt_reshape_479 = self.reshape_479(module20_18_opt, self.reshape_479_shape)
+ opt_split_480, opt_split_480_1 = self.split_480(opt_reshape_479)
+ opt_mul_481 = P.Mul()(opt_split_480, opt_split_467)
+ opt_mul_482 = P.Mul()(opt_split_480_1, opt_split_467_1)
+ opt_add_483 = opt_mul_481 + self.add_483_bias
+ opt_add_484 = P.Add()(opt_add_483, opt_mul_482)
+ opt_conv2d_485 = self.conv2d_485(opt_add_484)
+ opt_add_486 = P.Add()(opt_conv2d_485, opt_relu_462)
+ opt_relu_487 = self.relu_487(opt_add_486)
+ module22_20_opt = self.module22_20(opt_relu_487)
+ opt_split_492, opt_split_492_1 = self.split_492(module22_20_opt)
+ opt_add_493 = opt_split_492 + self.add_493_bias
+ opt_add_494 = P.Add()(opt_add_493, opt_split_492_1)
+ module20_19_opt = self.module20_19(opt_add_494)
+ opt_reshape_504 = self.reshape_504(module20_19_opt, self.reshape_504_shape)
+ opt_split_505, opt_split_505_1 = self.split_505(opt_reshape_504)
+ opt_mul_506 = P.Mul()(opt_split_505, opt_split_492)
+ opt_mul_507 = P.Mul()(opt_split_505_1, opt_split_492_1)
+ opt_add_508 = opt_mul_506 + self.add_508_bias
+ opt_add_509 = P.Add()(opt_add_508, opt_mul_507)
+ opt_conv2d_510 = self.conv2d_510(opt_add_509)
+ opt_add_511 = P.Add()(opt_conv2d_510, opt_relu_487)
+ opt_relu_512 = self.relu_512(opt_add_511)
+ module22_21_opt = self.module22_21(opt_relu_512)
+ opt_split_517, opt_split_517_1 = self.split_517(module22_21_opt)
+ opt_add_518 = opt_split_517 + self.add_518_bias
+ opt_add_519 = P.Add()(opt_add_518, opt_split_517_1)
+ module20_20_opt = self.module20_20(opt_add_519)
+ opt_reshape_529 = self.reshape_529(module20_20_opt, self.reshape_529_shape)
+ opt_split_530, opt_split_530_1 = self.split_530(opt_reshape_529)
+ opt_mul_531 = P.Mul()(opt_split_530, opt_split_517)
+ opt_mul_532 = P.Mul()(opt_split_530_1, opt_split_517_1)
+ opt_add_533 = opt_mul_531 + self.add_533_bias
+ opt_add_534 = P.Add()(opt_add_533, opt_mul_532)
+ opt_conv2d_535 = self.conv2d_535(opt_add_534)
+ opt_add_536 = P.Add()(opt_conv2d_535, opt_relu_512)
+ opt_relu_537 = self.relu_537(opt_add_536)
+ module22_22_opt = self.module22_22(opt_relu_537)
+ opt_split_542, opt_split_542_1 = self.split_542(module22_22_opt)
+ opt_add_543 = opt_split_542 + self.add_543_bias
+ opt_add_544 = P.Add()(opt_add_543, opt_split_542_1)
+ module20_21_opt = self.module20_21(opt_add_544)
+ opt_reshape_554 = self.reshape_554(module20_21_opt, self.reshape_554_shape)
+ opt_split_555, opt_split_555_1 = self.split_555(opt_reshape_554)
+ opt_mul_556 = P.Mul()(opt_split_555, opt_split_542)
+ opt_mul_557 = P.Mul()(opt_split_555_1, opt_split_542_1)
+ opt_add_558 = opt_mul_556 + self.add_558_bias
+ opt_add_559 = P.Add()(opt_add_558, opt_mul_557)
+ opt_conv2d_560 = self.conv2d_560(opt_add_559)
+ opt_add_561 = P.Add()(opt_conv2d_560, opt_relu_537)
+ opt_relu_562 = self.relu_562(opt_add_561)
+ module22_23_opt = self.module22_23(opt_relu_562)
+ opt_split_567, opt_split_567_1 = self.split_567(module22_23_opt)
+ opt_add_568 = opt_split_567 + self.add_568_bias
+ opt_add_569 = P.Add()(opt_add_568, opt_split_567_1)
+ module20_22_opt = self.module20_22(opt_add_569)
+ opt_reshape_579 = self.reshape_579(module20_22_opt, self.reshape_579_shape)
+ opt_split_580, opt_split_580_1 = self.split_580(opt_reshape_579)
+ opt_mul_581 = P.Mul()(opt_split_580, opt_split_567)
+ opt_mul_582 = P.Mul()(opt_split_580_1, opt_split_567_1)
+ opt_add_583 = opt_mul_581 + self.add_583_bias
+ opt_add_584 = P.Add()(opt_add_583, opt_mul_582)
+ opt_conv2d_585 = self.conv2d_585(opt_add_584)
+ opt_add_586 = P.Add()(opt_conv2d_585, opt_relu_562)
+ opt_relu_587 = self.relu_587(opt_add_586)
+ module22_24_opt = self.module22_24(opt_relu_587)
+ opt_split_592, opt_split_592_1 = self.split_592(module22_24_opt)
+ opt_add_593 = opt_split_592 + self.add_593_bias
+ opt_add_594 = P.Add()(opt_add_593, opt_split_592_1)
+ module20_23_opt = self.module20_23(opt_add_594)
+ opt_reshape_604 = self.reshape_604(module20_23_opt, self.reshape_604_shape)
+ opt_split_605, opt_split_605_1 = self.split_605(opt_reshape_604)
+ opt_mul_606 = P.Mul()(opt_split_605, opt_split_592)
+ opt_mul_607 = P.Mul()(opt_split_605_1, opt_split_592_1)
+ opt_add_608 = opt_mul_606 + self.add_608_bias
+ opt_add_609 = P.Add()(opt_add_608, opt_mul_607)
+ opt_conv2d_610 = self.conv2d_610(opt_add_609)
+ opt_add_611 = P.Add()(opt_conv2d_610, opt_relu_587)
+ opt_relu_612 = self.relu_612(opt_add_611)
+ module22_25_opt = self.module22_25(opt_relu_612)
+ opt_split_617, opt_split_617_1 = self.split_617(module22_25_opt)
+ opt_add_618 = opt_split_617 + self.add_618_bias
+ opt_add_619 = P.Add()(opt_add_618, opt_split_617_1)
+ module20_24_opt = self.module20_24(opt_add_619)
+ opt_reshape_629 = self.reshape_629(module20_24_opt, self.reshape_629_shape)
+ opt_split_630, opt_split_630_1 = self.split_630(opt_reshape_629)
+ opt_mul_631 = P.Mul()(opt_split_630, opt_split_617)
+ opt_mul_632 = P.Mul()(opt_split_630_1, opt_split_617_1)
+ opt_add_633 = opt_mul_631 + self.add_633_bias
+ opt_add_634 = P.Add()(opt_add_633, opt_mul_632)
+ opt_conv2d_635 = self.conv2d_635(opt_add_634)
+ opt_add_636 = P.Add()(opt_conv2d_635, opt_relu_612)
+ opt_relu_637 = self.relu_637(opt_add_636)
+ module22_26_opt = self.module22_26(opt_relu_637)
+ opt_split_642, opt_split_642_1 = self.split_642(module22_26_opt)
+ opt_add_643 = opt_split_642 + self.add_643_bias
+ opt_add_644 = P.Add()(opt_add_643, opt_split_642_1)
+ module20_25_opt = self.module20_25(opt_add_644)
+ opt_reshape_654 = self.reshape_654(module20_25_opt, self.reshape_654_shape)
+ opt_split_655, opt_split_655_1 = self.split_655(opt_reshape_654)
+ opt_mul_656 = P.Mul()(opt_split_655, opt_split_642)
+ opt_mul_657 = P.Mul()(opt_split_655_1, opt_split_642_1)
+ opt_add_658 = opt_mul_656 + self.add_658_bias
+ opt_add_659 = P.Add()(opt_add_658, opt_mul_657)
+ opt_conv2d_660 = self.conv2d_660(opt_add_659)
+ opt_add_661 = P.Add()(opt_conv2d_660, opt_relu_637)
+ opt_relu_662 = self.relu_662(opt_add_661)
+ module22_27_opt = self.module22_27(opt_relu_662)
+ opt_split_667, opt_split_667_1 = self.split_667(module22_27_opt)
+ opt_add_668 = opt_split_667 + self.add_668_bias
+ opt_add_669 = P.Add()(opt_add_668, opt_split_667_1)
+ module20_26_opt = self.module20_26(opt_add_669)
+ opt_reshape_679 = self.reshape_679(module20_26_opt, self.reshape_679_shape)
+ opt_split_680, opt_split_680_1 = self.split_680(opt_reshape_679)
+ opt_mul_681 = P.Mul()(opt_split_680, opt_split_667)
+ opt_mul_682 = P.Mul()(opt_split_680_1, opt_split_667_1)
+ opt_add_683 = opt_mul_681 + self.add_683_bias
+ opt_add_684 = P.Add()(opt_add_683, opt_mul_682)
+ opt_conv2d_685 = self.conv2d_685(opt_add_684)
+ opt_add_686 = P.Add()(opt_conv2d_685, opt_relu_662)
+ opt_relu_687 = self.relu_687(opt_add_686)
+ module22_28_opt = self.module22_28(opt_relu_687)
+ opt_split_694, opt_split_694_1 = self.split_694(module22_28_opt)
+ opt_add_695 = opt_split_694 + self.add_695_bias
+ opt_add_696 = P.Add()(opt_add_695, opt_split_694_1)
+ module20_27_opt = self.module20_27(opt_add_696)
+ opt_reshape_706 = self.reshape_706(module20_27_opt, self.reshape_706_shape)
+ opt_split_707, opt_split_707_1 = self.split_707(opt_reshape_706)
+ opt_mul_708 = P.Mul()(opt_split_707, opt_split_694)
+ opt_mul_709 = P.Mul()(opt_split_707_1, opt_split_694_1)
+ opt_add_710 = opt_mul_708 + self.add_710_bias
+ opt_add_711 = P.Add()(opt_add_710, opt_mul_709)
+ opt_pad_712 = self.pad_712(opt_add_711)
+ opt_avgpool2d_713 = self.pad_avgpool2d_713(opt_pad_712)
+ opt_avgpool2d_713 = self.avgpool2d_713(opt_avgpool2d_713)
+ opt_conv2d_714 = self.conv2d_714(opt_avgpool2d_713)
+ opt_avgpool2d_689 = self.pad_avgpool2d_689(opt_relu_687)
+ opt_avgpool2d_689 = self.avgpool2d_689(opt_avgpool2d_689)
+ opt_conv2d_691 = self.conv2d_691(opt_avgpool2d_689)
+ opt_add_715 = P.Add()(opt_conv2d_714, opt_conv2d_691)
+ opt_relu_716 = self.relu_716(opt_add_715)
+ module22_29_opt = self.module22_29(opt_relu_716)
+ opt_split_721, opt_split_721_1 = self.split_721(module22_29_opt)
+ opt_add_722 = opt_split_721 + self.add_722_bias
+ opt_add_723 = P.Add()(opt_add_722, opt_split_721_1)
+ module20_28_opt = self.module20_28(opt_add_723)
+ opt_reshape_733 = self.reshape_733(module20_28_opt, self.reshape_733_shape)
+ opt_split_734, opt_split_734_1 = self.split_734(opt_reshape_733)
+ opt_mul_735 = P.Mul()(opt_split_734, opt_split_721)
+ opt_mul_736 = P.Mul()(opt_split_734_1, opt_split_721_1)
+ opt_add_737 = opt_mul_735 + self.add_737_bias
+ opt_add_738 = P.Add()(opt_add_737, opt_mul_736)
+ opt_conv2d_739 = self.conv2d_739(opt_add_738)
+ opt_add_740 = P.Add()(opt_conv2d_739, opt_relu_716)
+ opt_relu_741 = self.relu_741(opt_add_740)
+ module22_30_opt = self.module22_30(opt_relu_741)
+ opt_split_746, opt_split_746_1 = self.split_746(module22_30_opt)
+ opt_add_747 = opt_split_746 + self.add_747_bias
+ opt_add_748 = P.Add()(opt_add_747, opt_split_746_1)
+ module20_29_opt = self.module20_29(opt_add_748)
+ opt_reshape_758 = self.reshape_758(module20_29_opt, self.reshape_758_shape)
+ opt_split_759, opt_split_759_1 = self.split_759(opt_reshape_758)
+ opt_mul_760 = P.Mul()(opt_split_759, opt_split_746)
+ opt_mul_761 = P.Mul()(opt_split_759_1, opt_split_746_1)
+ opt_add_762 = opt_mul_760 + self.add_762_bias
+ opt_add_763 = P.Add()(opt_add_762, opt_mul_761)
+ opt_conv2d_764 = self.conv2d_764(opt_add_763)
+ opt_add_765 = P.Add()(opt_conv2d_764, opt_relu_741)
+ opt_relu_766 = self.relu_766(opt_add_765)
+ module22_31_opt = self.module22_31(opt_relu_766)
+ opt_split_771, opt_split_771_1 = self.split_771(module22_31_opt)
+ opt_add_772 = opt_split_771 + self.add_772_bias
+ opt_add_773 = P.Add()(opt_add_772, opt_split_771_1)
+ module20_30_opt = self.module20_30(opt_add_773)
+ opt_reshape_783 = self.reshape_783(module20_30_opt, self.reshape_783_shape)
+ opt_split_784, opt_split_784_1 = self.split_784(opt_reshape_783)
+ opt_mul_785 = P.Mul()(opt_split_784, opt_split_771)
+ opt_mul_786 = P.Mul()(opt_split_784_1, opt_split_771_1)
+ opt_add_787 = opt_mul_785 + self.add_787_bias
+ opt_add_788 = P.Add()(opt_add_787, opt_mul_786)
+ opt_conv2d_789 = self.conv2d_789(opt_add_788)
+ opt_add_790 = P.Add()(opt_conv2d_789, opt_relu_766)
+ opt_relu_791 = self.relu_791(opt_add_790)
+ module22_32_opt = self.module22_32(opt_relu_791)
+ opt_split_796, opt_split_796_1 = self.split_796(module22_32_opt)
+ opt_add_797 = opt_split_796 + self.add_797_bias
+ opt_add_798 = P.Add()(opt_add_797, opt_split_796_1)
+ module20_31_opt = self.module20_31(opt_add_798)
+ opt_reshape_808 = self.reshape_808(module20_31_opt, self.reshape_808_shape)
+ opt_split_809, opt_split_809_1 = self.split_809(opt_reshape_808)
+ opt_mul_810 = P.Mul()(opt_split_809, opt_split_796)
+ opt_mul_811 = P.Mul()(opt_split_809_1, opt_split_796_1)
+ opt_add_812 = opt_mul_810 + self.add_812_bias
+ opt_add_813 = P.Add()(opt_add_812, opt_mul_811)
+ opt_conv2d_814 = self.conv2d_814(opt_add_813)
+ opt_add_815 = P.Add()(opt_conv2d_814, opt_relu_791)
+ opt_relu_816 = self.relu_816(opt_add_815)
+ module22_33_opt = self.module22_33(opt_relu_816)
+ opt_split_821, opt_split_821_1 = self.split_821(module22_33_opt)
+ opt_add_822 = opt_split_821 + self.add_822_bias
+ opt_add_823 = P.Add()(opt_add_822, opt_split_821_1)
+ module20_32_opt = self.module20_32(opt_add_823)
+ opt_reshape_833 = self.reshape_833(module20_32_opt, self.reshape_833_shape)
+ opt_split_834, opt_split_834_1 = self.split_834(opt_reshape_833)
+ opt_mul_835 = P.Mul()(opt_split_834, opt_split_821)
+ opt_mul_836 = P.Mul()(opt_split_834_1, opt_split_821_1)
+ opt_add_837 = opt_mul_835 + self.add_837_bias
+ opt_add_838 = P.Add()(opt_add_837, opt_mul_836)
+ opt_conv2d_839 = self.conv2d_839(opt_add_838)
+ opt_add_840 = P.Add()(opt_conv2d_839, opt_relu_816)
+ opt_relu_841 = self.relu_841(opt_add_840)
+ module22_34_opt = self.module22_34(opt_relu_841)
+ opt_split_846, opt_split_846_1 = self.split_846(module22_34_opt)
+ opt_add_847 = opt_split_846 + self.add_847_bias
+ opt_add_848 = P.Add()(opt_add_847, opt_split_846_1)
+ module20_33_opt = self.module20_33(opt_add_848)
+ opt_reshape_858 = self.reshape_858(module20_33_opt, self.reshape_858_shape)
+ opt_split_859, opt_split_859_1 = self.split_859(opt_reshape_858)
+ opt_mul_860 = P.Mul()(opt_split_859, opt_split_846)
+ opt_mul_861 = P.Mul()(opt_split_859_1, opt_split_846_1)
+ opt_add_862 = opt_mul_860 + self.add_862_bias
+ opt_add_863 = P.Add()(opt_add_862, opt_mul_861)
+ opt_conv2d_864 = self.conv2d_864(opt_add_863)
+ opt_add_865 = P.Add()(opt_conv2d_864, opt_relu_841)
+ opt_relu_866 = self.relu_866(opt_add_865)
+ module22_35_opt = self.module22_35(opt_relu_866)
+ opt_split_871, opt_split_871_1 = self.split_871(module22_35_opt)
+ opt_add_872 = opt_split_871 + self.add_872_bias
+ opt_add_873 = P.Add()(opt_add_872, opt_split_871_1)
+ module20_34_opt = self.module20_34(opt_add_873)
+ opt_reshape_883 = self.reshape_883(module20_34_opt, self.reshape_883_shape)
+ opt_split_884, opt_split_884_1 = self.split_884(opt_reshape_883)
+ opt_mul_885 = P.Mul()(opt_split_884, opt_split_871)
+ opt_mul_886 = P.Mul()(opt_split_884_1, opt_split_871_1)
+ opt_add_887 = opt_mul_885 + self.add_887_bias
+ opt_add_888 = P.Add()(opt_add_887, opt_mul_886)
+ opt_conv2d_889 = self.conv2d_889(opt_add_888)
+ opt_add_890 = P.Add()(opt_conv2d_889, opt_relu_866)
+ opt_relu_891 = self.relu_891(opt_add_890)
+ module22_36_opt = self.module22_36(opt_relu_891)
+ opt_split_896, opt_split_896_1 = self.split_896(module22_36_opt)
+ opt_add_897 = opt_split_896 + self.add_897_bias
+ opt_add_898 = P.Add()(opt_add_897, opt_split_896_1)
+ module20_35_opt = self.module20_35(opt_add_898)
+ opt_reshape_908 = self.reshape_908(module20_35_opt, self.reshape_908_shape)
+ opt_split_909, opt_split_909_1 = self.split_909(opt_reshape_908)
+ opt_mul_910 = P.Mul()(opt_split_909, opt_split_896)
+ opt_mul_911 = P.Mul()(opt_split_909_1, opt_split_896_1)
+ opt_add_912 = opt_mul_910 + self.add_912_bias
+ opt_add_913 = P.Add()(opt_add_912, opt_mul_911)
+ opt_conv2d_914 = self.conv2d_914(opt_add_913)
+ opt_add_915 = P.Add()(opt_conv2d_914, opt_relu_891)
+ opt_relu_916 = self.relu_916(opt_add_915)
+ module22_37_opt = self.module22_37(opt_relu_916)
+ opt_split_921, opt_split_921_1 = self.split_921(module22_37_opt)
+ opt_add_922 = opt_split_921 + self.add_922_bias
+ opt_add_923 = P.Add()(opt_add_922, opt_split_921_1)
+ module20_36_opt = self.module20_36(opt_add_923)
+ opt_reshape_933 = self.reshape_933(module20_36_opt, self.reshape_933_shape)
+ opt_split_934, opt_split_934_1 = self.split_934(opt_reshape_933)
+ opt_mul_935 = P.Mul()(opt_split_934, opt_split_921)
+ opt_mul_936 = P.Mul()(opt_split_934_1, opt_split_921_1)
+ opt_add_937 = opt_mul_935 + self.add_937_bias
+ opt_add_938 = P.Add()(opt_add_937, opt_mul_936)
+ opt_conv2d_939 = self.conv2d_939(opt_add_938)
+ opt_add_940 = P.Add()(opt_conv2d_939, opt_relu_916)
+ opt_relu_941 = self.relu_941(opt_add_940)
+ module22_38_opt = self.module22_38(opt_relu_941)
+ opt_split_946, opt_split_946_1 = self.split_946(module22_38_opt)
+ opt_add_947 = opt_split_946 + self.add_947_bias
+ opt_add_948 = P.Add()(opt_add_947, opt_split_946_1)
+ module20_37_opt = self.module20_37(opt_add_948)
+ opt_reshape_958 = self.reshape_958(module20_37_opt, self.reshape_958_shape)
+ opt_split_959, opt_split_959_1 = self.split_959(opt_reshape_958)
+ opt_mul_960 = P.Mul()(opt_split_959, opt_split_946)
+ opt_mul_961 = P.Mul()(opt_split_959_1, opt_split_946_1)
+ opt_add_962 = opt_mul_960 + self.add_962_bias
+ opt_add_963 = P.Add()(opt_add_962, opt_mul_961)
+ opt_conv2d_964 = self.conv2d_964(opt_add_963)
+ opt_add_965 = P.Add()(opt_conv2d_964, opt_relu_941)
+ opt_relu_966 = self.relu_966(opt_add_965)
+ module22_39_opt = self.module22_39(opt_relu_966)
+ opt_split_971, opt_split_971_1 = self.split_971(module22_39_opt)
+ opt_add_972 = opt_split_971 + self.add_972_bias
+ opt_add_973 = P.Add()(opt_add_972, opt_split_971_1)
+ module20_38_opt = self.module20_38(opt_add_973)
+ opt_reshape_983 = self.reshape_983(module20_38_opt, self.reshape_983_shape)
+ opt_split_984, opt_split_984_1 = self.split_984(opt_reshape_983)
+ opt_mul_985 = P.Mul()(opt_split_984, opt_split_971)
+ opt_mul_986 = P.Mul()(opt_split_984_1, opt_split_971_1)
+ opt_add_987 = opt_mul_985 + self.add_987_bias
+ opt_add_988 = P.Add()(opt_add_987, opt_mul_986)
+ opt_conv2d_989 = self.conv2d_989(opt_add_988)
+ opt_add_990 = P.Add()(opt_conv2d_989, opt_relu_966)
+ opt_relu_991 = self.relu_991(opt_add_990)
+ module22_40_opt = self.module22_40(opt_relu_991)
+ opt_split_996, opt_split_996_1 = self.split_996(module22_40_opt)
+ opt_add_997 = opt_split_996 + self.add_997_bias
+ opt_add_998 = P.Add()(opt_add_997, opt_split_996_1)
+ module20_39_opt = self.module20_39(opt_add_998)
+ opt_reshape_1008 = self.reshape_1008(module20_39_opt, self.reshape_1008_shape)
+ opt_split_1009, opt_split_1009_1 = self.split_1009(opt_reshape_1008)
+ opt_mul_1010 = P.Mul()(opt_split_1009, opt_split_996)
+ opt_mul_1011 = P.Mul()(opt_split_1009_1, opt_split_996_1)
+ opt_add_1012 = opt_mul_1010 + self.add_1012_bias
+ opt_add_1013 = P.Add()(opt_add_1012, opt_mul_1011)
+ opt_conv2d_1014 = self.conv2d_1014(opt_add_1013)
+ opt_add_1015 = P.Add()(opt_conv2d_1014, opt_relu_991)
+ opt_relu_1016 = self.relu_1016(opt_add_1015)
+ module22_41_opt = self.module22_41(opt_relu_1016)
+ opt_split_1021, opt_split_1021_1 = self.split_1021(module22_41_opt)
+ opt_add_1022 = opt_split_1021 + self.add_1022_bias
+ opt_add_1023 = P.Add()(opt_add_1022, opt_split_1021_1)
+ module20_40_opt = self.module20_40(opt_add_1023)
+ opt_reshape_1033 = self.reshape_1033(module20_40_opt, self.reshape_1033_shape)
+ opt_split_1034, opt_split_1034_1 = self.split_1034(opt_reshape_1033)
+ opt_mul_1035 = P.Mul()(opt_split_1034, opt_split_1021)
+ opt_mul_1036 = P.Mul()(opt_split_1034_1, opt_split_1021_1)
+ opt_add_1037 = opt_mul_1035 + self.add_1037_bias
+ opt_add_1038 = P.Add()(opt_add_1037, opt_mul_1036)
+ opt_conv2d_1039 = self.conv2d_1039(opt_add_1038)
+ opt_add_1040 = P.Add()(opt_conv2d_1039, opt_relu_1016)
+ opt_relu_1041 = self.relu_1041(opt_add_1040)
+ module22_42_opt = self.module22_42(opt_relu_1041)
+ opt_split_1046, opt_split_1046_1 = self.split_1046(module22_42_opt)
+ opt_add_1047 = opt_split_1046 + self.add_1047_bias
+ opt_add_1048 = P.Add()(opt_add_1047, opt_split_1046_1)
+ module20_41_opt = self.module20_41(opt_add_1048)
+ opt_reshape_1058 = self.reshape_1058(module20_41_opt, self.reshape_1058_shape)
+ opt_split_1059, opt_split_1059_1 = self.split_1059(opt_reshape_1058)
+ opt_mul_1060 = P.Mul()(opt_split_1059, opt_split_1046)
+ opt_mul_1061 = P.Mul()(opt_split_1059_1, opt_split_1046_1)
+ opt_add_1062 = opt_mul_1060 + self.add_1062_bias
+ opt_add_1063 = P.Add()(opt_add_1062, opt_mul_1061)
+ opt_conv2d_1064 = self.conv2d_1064(opt_add_1063)
+ opt_add_1065 = P.Add()(opt_conv2d_1064, opt_relu_1041)
+ opt_relu_1066 = self.relu_1066(opt_add_1065)
+ module22_43_opt = self.module22_43(opt_relu_1066)
+ opt_split_1071, opt_split_1071_1 = self.split_1071(module22_43_opt)
+ opt_add_1072 = opt_split_1071 + self.add_1072_bias
+ opt_add_1073 = P.Add()(opt_add_1072, opt_split_1071_1)
+ module20_42_opt = self.module20_42(opt_add_1073)
+ opt_reshape_1083 = self.reshape_1083(module20_42_opt, self.reshape_1083_shape)
+ opt_split_1084, opt_split_1084_1 = self.split_1084(opt_reshape_1083)
+ opt_mul_1085 = P.Mul()(opt_split_1084, opt_split_1071)
+ opt_mul_1086 = P.Mul()(opt_split_1084_1, opt_split_1071_1)
+ opt_add_1087 = opt_mul_1085 + self.add_1087_bias
+ opt_add_1088 = P.Add()(opt_add_1087, opt_mul_1086)
+ opt_conv2d_1089 = self.conv2d_1089(opt_add_1088)
+ opt_add_1090 = P.Add()(opt_conv2d_1089, opt_relu_1066)
+ opt_relu_1091 = self.relu_1091(opt_add_1090)
+ module22_44_opt = self.module22_44(opt_relu_1091)
+ opt_split_1096, opt_split_1096_1 = self.split_1096(module22_44_opt)
+ opt_add_1097 = opt_split_1096 + self.add_1097_bias
+ opt_add_1098 = P.Add()(opt_add_1097, opt_split_1096_1)
+ module20_43_opt = self.module20_43(opt_add_1098)
+ opt_reshape_1108 = self.reshape_1108(module20_43_opt, self.reshape_1108_shape)
+ opt_split_1109, opt_split_1109_1 = self.split_1109(opt_reshape_1108)
+ opt_mul_1110 = P.Mul()(opt_split_1109, opt_split_1096)
+ opt_mul_1111 = P.Mul()(opt_split_1109_1, opt_split_1096_1)
+ opt_add_1112 = opt_mul_1110 + self.add_1112_bias
+ opt_add_1113 = P.Add()(opt_add_1112, opt_mul_1111)
+ opt_conv2d_1114 = self.conv2d_1114(opt_add_1113)
+ opt_add_1115 = P.Add()(opt_conv2d_1114, opt_relu_1091)
+ opt_relu_1116 = self.relu_1116(opt_add_1115)
+ module22_45_opt = self.module22_45(opt_relu_1116)
+ opt_split_1121, opt_split_1121_1 = self.split_1121(module22_45_opt)
+ opt_add_1122 = opt_split_1121 + self.add_1122_bias
+ opt_add_1123 = P.Add()(opt_add_1122, opt_split_1121_1)
+ module20_44_opt = self.module20_44(opt_add_1123)
+ opt_reshape_1133 = self.reshape_1133(module20_44_opt, self.reshape_1133_shape)
+ opt_split_1134, opt_split_1134_1 = self.split_1134(opt_reshape_1133)
+ opt_mul_1135 = P.Mul()(opt_split_1134, opt_split_1121)
+ opt_mul_1136 = P.Mul()(opt_split_1134_1, opt_split_1121_1)
+ opt_add_1137 = opt_mul_1135 + self.add_1137_bias
+ opt_add_1138 = P.Add()(opt_add_1137, opt_mul_1136)
+ opt_conv2d_1139 = self.conv2d_1139(opt_add_1138)
+ opt_add_1140 = P.Add()(opt_conv2d_1139, opt_relu_1116)
+ opt_relu_1141 = self.relu_1141(opt_add_1140)
+ module22_46_opt = self.module22_46(opt_relu_1141)
+ opt_split_1146, opt_split_1146_1 = self.split_1146(module22_46_opt)
+ opt_add_1147 = opt_split_1146 + self.add_1147_bias
+ opt_add_1148 = P.Add()(opt_add_1147, opt_split_1146_1)
+ module20_45_opt = self.module20_45(opt_add_1148)
+ opt_reshape_1158 = self.reshape_1158(module20_45_opt, self.reshape_1158_shape)
+ opt_split_1159, opt_split_1159_1 = self.split_1159(opt_reshape_1158)
+ opt_mul_1160 = P.Mul()(opt_split_1159, opt_split_1146)
+ opt_mul_1161 = P.Mul()(opt_split_1159_1, opt_split_1146_1)
+ opt_add_1162 = opt_mul_1160 + self.add_1162_bias
+ opt_add_1163 = P.Add()(opt_add_1162, opt_mul_1161)
+ opt_conv2d_1164 = self.conv2d_1164(opt_add_1163)
+ opt_add_1165 = P.Add()(opt_conv2d_1164, opt_relu_1141)
+ opt_relu_1166 = self.relu_1166(opt_add_1165)
+ module22_47_opt = self.module22_47(opt_relu_1166)
+ opt_split_1171, opt_split_1171_1 = self.split_1171(module22_47_opt)
+ opt_add_1172 = opt_split_1171 + self.add_1172_bias
+ opt_add_1173 = P.Add()(opt_add_1172, opt_split_1171_1)
+ module20_46_opt = self.module20_46(opt_add_1173)
+ opt_reshape_1183 = self.reshape_1183(module20_46_opt, self.reshape_1183_shape)
+ opt_split_1184, opt_split_1184_1 = self.split_1184(opt_reshape_1183)
+ opt_mul_1185 = P.Mul()(opt_split_1184, opt_split_1171)
+ opt_mul_1186 = P.Mul()(opt_split_1184_1, opt_split_1171_1)
+ opt_add_1187 = opt_mul_1185 + self.add_1187_bias
+ opt_add_1188 = P.Add()(opt_add_1187, opt_mul_1186)
+ opt_conv2d_1189 = self.conv2d_1189(opt_add_1188)
+ opt_add_1190 = P.Add()(opt_conv2d_1189, opt_relu_1166)
+ opt_relu_1191 = self.relu_1191(opt_add_1190)
+ module22_48_opt = self.module22_48(opt_relu_1191)
+ opt_split_1196, opt_split_1196_1 = self.split_1196(module22_48_opt)
+ opt_add_1197 = opt_split_1196 + self.add_1197_bias
+ opt_add_1198 = P.Add()(opt_add_1197, opt_split_1196_1)
+ module20_47_opt = self.module20_47(opt_add_1198)
+ opt_reshape_1208 = self.reshape_1208(module20_47_opt, self.reshape_1208_shape)
+ opt_split_1209, opt_split_1209_1 = self.split_1209(opt_reshape_1208)
+ opt_mul_1210 = P.Mul()(opt_split_1209, opt_split_1196)
+ opt_mul_1211 = P.Mul()(opt_split_1209_1, opt_split_1196_1)
+ opt_add_1212 = opt_mul_1210 + self.add_1212_bias
+ opt_add_1213 = P.Add()(opt_add_1212, opt_mul_1211)
+ opt_conv2d_1214 = self.conv2d_1214(opt_add_1213)
+ opt_add_1215 = P.Add()(opt_conv2d_1214, opt_relu_1191)
+ opt_relu_1216 = self.relu_1216(opt_add_1215)
+ module22_49_opt = self.module22_49(opt_relu_1216)
+ opt_split_1221, opt_split_1221_1 = self.split_1221(module22_49_opt)
+ opt_add_1222 = opt_split_1221 + self.add_1222_bias
+ opt_add_1223 = P.Add()(opt_add_1222, opt_split_1221_1)
+ module20_48_opt = self.module20_48(opt_add_1223)
+ opt_reshape_1233 = self.reshape_1233(module20_48_opt, self.reshape_1233_shape)
+ opt_split_1234, opt_split_1234_1 = self.split_1234(opt_reshape_1233)
+ opt_mul_1235 = P.Mul()(opt_split_1234, opt_split_1221)
+ opt_mul_1236 = P.Mul()(opt_split_1234_1, opt_split_1221_1)
+ opt_add_1237 = opt_mul_1235 + self.add_1237_bias
+ opt_add_1238 = P.Add()(opt_add_1237, opt_mul_1236)
+ opt_conv2d_1239 = self.conv2d_1239(opt_add_1238)
+ opt_add_1240 = P.Add()(opt_conv2d_1239, opt_relu_1216)
+ opt_relu_1241 = self.relu_1241(opt_add_1240)
+ module22_50_opt = self.module22_50(opt_relu_1241)
+ opt_split_1246, opt_split_1246_1 = self.split_1246(module22_50_opt)
+ opt_add_1247 = opt_split_1246 + self.add_1247_bias
+ opt_add_1248 = P.Add()(opt_add_1247, opt_split_1246_1)
+ module20_49_opt = self.module20_49(opt_add_1248)
+ opt_reshape_1258 = self.reshape_1258(module20_49_opt, self.reshape_1258_shape)
+ opt_split_1259, opt_split_1259_1 = self.split_1259(opt_reshape_1258)
+ opt_mul_1260 = P.Mul()(opt_split_1259, opt_split_1246)
+ opt_mul_1261 = P.Mul()(opt_split_1259_1, opt_split_1246_1)
+ opt_add_1262 = opt_mul_1260 + self.add_1262_bias
+ opt_add_1263 = P.Add()(opt_add_1262, opt_mul_1261)
+ opt_conv2d_1264 = self.conv2d_1264(opt_add_1263)
+ opt_add_1265 = P.Add()(opt_conv2d_1264, opt_relu_1241)
+ opt_relu_1266 = self.relu_1266(opt_add_1265)
+ module22_51_opt = self.module22_51(opt_relu_1266)
+ opt_split_1271, opt_split_1271_1 = self.split_1271(module22_51_opt)
+ opt_add_1272 = opt_split_1271 + self.add_1272_bias
+ opt_add_1273 = P.Add()(opt_add_1272, opt_split_1271_1)
+ module20_50_opt = self.module20_50(opt_add_1273)
+ opt_reshape_1283 = self.reshape_1283(module20_50_opt, self.reshape_1283_shape)
+ opt_split_1284, opt_split_1284_1 = self.split_1284(opt_reshape_1283)
+ opt_mul_1285 = P.Mul()(opt_split_1284, opt_split_1271)
+ opt_mul_1286 = P.Mul()(opt_split_1284_1, opt_split_1271_1)
+ opt_add_1287 = opt_mul_1285 + self.add_1287_bias
+ opt_add_1288 = P.Add()(opt_add_1287, opt_mul_1286)
+ opt_conv2d_1289 = self.conv2d_1289(opt_add_1288)
+ opt_add_1290 = P.Add()(opt_conv2d_1289, opt_relu_1266)
+ opt_relu_1291 = self.relu_1291(opt_add_1290)
+ module22_52_opt = self.module22_52(opt_relu_1291)
+ opt_split_1296, opt_split_1296_1 = self.split_1296(module22_52_opt)
+ opt_add_1297 = opt_split_1296 + self.add_1297_bias
+ opt_add_1298 = P.Add()(opt_add_1297, opt_split_1296_1)
+ module20_51_opt = self.module20_51(opt_add_1298)
+ opt_reshape_1308 = self.reshape_1308(module20_51_opt, self.reshape_1308_shape)
+ opt_split_1309, opt_split_1309_1 = self.split_1309(opt_reshape_1308)
+ opt_mul_1310 = P.Mul()(opt_split_1309, opt_split_1296)
+ opt_mul_1311 = P.Mul()(opt_split_1309_1, opt_split_1296_1)
+ opt_add_1312 = opt_mul_1310 + self.add_1312_bias
+ opt_add_1313 = P.Add()(opt_add_1312, opt_mul_1311)
+ opt_conv2d_1314 = self.conv2d_1314(opt_add_1313)
+ opt_add_1315 = P.Add()(opt_conv2d_1314, opt_relu_1291)
+ opt_relu_1316 = self.relu_1316(opt_add_1315)
+ module22_53_opt = self.module22_53(opt_relu_1316)
+ opt_split_1321, opt_split_1321_1 = self.split_1321(module22_53_opt)
+ opt_add_1322 = opt_split_1321 + self.add_1322_bias
+ opt_add_1323 = P.Add()(opt_add_1322, opt_split_1321_1)
+ module20_52_opt = self.module20_52(opt_add_1323)
+ opt_reshape_1333 = self.reshape_1333(module20_52_opt, self.reshape_1333_shape)
+ opt_split_1334, opt_split_1334_1 = self.split_1334(opt_reshape_1333)
+ opt_mul_1335 = P.Mul()(opt_split_1334, opt_split_1321)
+ opt_mul_1336 = P.Mul()(opt_split_1334_1, opt_split_1321_1)
+ opt_add_1337 = opt_mul_1335 + self.add_1337_bias
+ opt_add_1338 = P.Add()(opt_add_1337, opt_mul_1336)
+ opt_conv2d_1339 = self.conv2d_1339(opt_add_1338)
+ opt_add_1340 = P.Add()(opt_conv2d_1339, opt_relu_1316)
+ opt_relu_1341 = self.relu_1341(opt_add_1340)
+ module22_54_opt = self.module22_54(opt_relu_1341)
+ opt_split_1346, opt_split_1346_1 = self.split_1346(module22_54_opt)
+ opt_add_1347 = opt_split_1346 + self.add_1347_bias
+ opt_add_1348 = P.Add()(opt_add_1347, opt_split_1346_1)
+ module20_53_opt = self.module20_53(opt_add_1348)
+ opt_reshape_1358 = self.reshape_1358(module20_53_opt, self.reshape_1358_shape)
+ opt_split_1359, opt_split_1359_1 = self.split_1359(opt_reshape_1358)
+ opt_mul_1360 = P.Mul()(opt_split_1359, opt_split_1346)
+ opt_mul_1361 = P.Mul()(opt_split_1359_1, opt_split_1346_1)
+ opt_add_1362 = opt_mul_1360 + self.add_1362_bias
+ opt_add_1363 = P.Add()(opt_add_1362, opt_mul_1361)
+ opt_conv2d_1364 = self.conv2d_1364(opt_add_1363)
+ opt_add_1365 = P.Add()(opt_conv2d_1364, opt_relu_1341)
+ opt_relu_1366 = self.relu_1366(opt_add_1365)
+ module22_55_opt = self.module22_55(opt_relu_1366)
+ opt_split_1371, opt_split_1371_1 = self.split_1371(module22_55_opt)
+ opt_add_1372 = opt_split_1371 + self.add_1372_bias
+ opt_add_1373 = P.Add()(opt_add_1372, opt_split_1371_1)
+ module20_54_opt = self.module20_54(opt_add_1373)
+ opt_reshape_1383 = self.reshape_1383(module20_54_opt, self.reshape_1383_shape)
+ opt_split_1384, opt_split_1384_1 = self.split_1384(opt_reshape_1383)
+ opt_mul_1385 = P.Mul()(opt_split_1384, opt_split_1371)
+ opt_mul_1386 = P.Mul()(opt_split_1384_1, opt_split_1371_1)
+ opt_add_1387 = opt_mul_1385 + self.add_1387_bias
+ opt_add_1388 = P.Add()(opt_add_1387, opt_mul_1386)
+ opt_conv2d_1389 = self.conv2d_1389(opt_add_1388)
+ opt_add_1390 = P.Add()(opt_conv2d_1389, opt_relu_1366)
+ opt_relu_1391 = self.relu_1391(opt_add_1390)
+ module22_56_opt = self.module22_56(opt_relu_1391)
+ opt_split_1396, opt_split_1396_1 = self.split_1396(module22_56_opt)
+ opt_add_1397 = opt_split_1396 + self.add_1397_bias
+ opt_add_1398 = P.Add()(opt_add_1397, opt_split_1396_1)
+ module20_55_opt = self.module20_55(opt_add_1398)
+ opt_reshape_1408 = self.reshape_1408(module20_55_opt, self.reshape_1408_shape)
+ opt_split_1409, opt_split_1409_1 = self.split_1409(opt_reshape_1408)
+ opt_mul_1410 = P.Mul()(opt_split_1409, opt_split_1396)
+ opt_mul_1411 = P.Mul()(opt_split_1409_1, opt_split_1396_1)
+ opt_add_1412 = opt_mul_1410 + self.add_1412_bias
+ opt_add_1413 = P.Add()(opt_add_1412, opt_mul_1411)
+ opt_conv2d_1414 = self.conv2d_1414(opt_add_1413)
+ opt_add_1415 = P.Add()(opt_conv2d_1414, opt_relu_1391)
+ opt_relu_1416 = self.relu_1416(opt_add_1415)
+ module22_57_opt = self.module22_57(opt_relu_1416)
+ opt_split_1421, opt_split_1421_1 = self.split_1421(module22_57_opt)
+ opt_add_1422 = opt_split_1421 + self.add_1422_bias
+ opt_add_1423 = P.Add()(opt_add_1422, opt_split_1421_1)
+ module20_56_opt = self.module20_56(opt_add_1423)
+ opt_reshape_1433 = self.reshape_1433(module20_56_opt, self.reshape_1433_shape)
+ opt_split_1434, opt_split_1434_1 = self.split_1434(opt_reshape_1433)
+ opt_mul_1435 = P.Mul()(opt_split_1434, opt_split_1421)
+ opt_mul_1436 = P.Mul()(opt_split_1434_1, opt_split_1421_1)
+ opt_add_1437 = opt_mul_1435 + self.add_1437_bias
+ opt_add_1438 = P.Add()(opt_add_1437, opt_mul_1436)
+ opt_conv2d_1439 = self.conv2d_1439(opt_add_1438)
+ opt_add_1440 = P.Add()(opt_conv2d_1439, opt_relu_1416)
+ opt_relu_1441 = self.relu_1441(opt_add_1440)
+ module22_58_opt = self.module22_58(opt_relu_1441)
+ opt_split_1446, opt_split_1446_1 = self.split_1446(module22_58_opt)
+ opt_add_1447 = opt_split_1446 + self.add_1447_bias
+ opt_add_1448 = P.Add()(opt_add_1447, opt_split_1446_1)
+ module20_57_opt = self.module20_57(opt_add_1448)
+ opt_reshape_1458 = self.reshape_1458(module20_57_opt, self.reshape_1458_shape)
+ opt_split_1459, opt_split_1459_1 = self.split_1459(opt_reshape_1458)
+ opt_mul_1460 = P.Mul()(opt_split_1459, opt_split_1446)
+ opt_mul_1461 = P.Mul()(opt_split_1459_1, opt_split_1446_1)
+ opt_add_1462 = opt_mul_1460 + self.add_1462_bias
+ opt_add_1463 = P.Add()(opt_add_1462, opt_mul_1461)
+ opt_conv2d_1464 = self.conv2d_1464(opt_add_1463)
+ opt_add_1465 = P.Add()(opt_conv2d_1464, opt_relu_1441)
+ opt_relu_1466 = self.relu_1466(opt_add_1465)
+ module22_59_opt = self.module22_59(opt_relu_1466)
+ opt_split_1471, opt_split_1471_1 = self.split_1471(module22_59_opt)
+ opt_add_1472 = opt_split_1471 + self.add_1472_bias
+ opt_add_1473 = P.Add()(opt_add_1472, opt_split_1471_1)
+ module20_58_opt = self.module20_58(opt_add_1473)
+ opt_reshape_1483 = self.reshape_1483(module20_58_opt, self.reshape_1483_shape)
+ opt_split_1484, opt_split_1484_1 = self.split_1484(opt_reshape_1483)
+ opt_mul_1485 = P.Mul()(opt_split_1484, opt_split_1471)
+ opt_mul_1486 = P.Mul()(opt_split_1484_1, opt_split_1471_1)
+ opt_add_1487 = opt_mul_1485 + self.add_1487_bias
+ opt_add_1488 = P.Add()(opt_add_1487, opt_mul_1486)
+ opt_conv2d_1489 = self.conv2d_1489(opt_add_1488)
+ opt_add_1490 = P.Add()(opt_conv2d_1489, opt_relu_1466)
+ opt_relu_1491 = self.relu_1491(opt_add_1490)
+ module22_60_opt = self.module22_60(opt_relu_1491)
+ opt_split_1496, opt_split_1496_1 = self.split_1496(module22_60_opt)
+ opt_add_1497 = opt_split_1496 + self.add_1497_bias
+ opt_add_1498 = P.Add()(opt_add_1497, opt_split_1496_1)
+ module20_59_opt = self.module20_59(opt_add_1498)
+ opt_reshape_1508 = self.reshape_1508(module20_59_opt, self.reshape_1508_shape)
+ opt_split_1509, opt_split_1509_1 = self.split_1509(opt_reshape_1508)
+ opt_mul_1510 = P.Mul()(opt_split_1509, opt_split_1496)
+ opt_mul_1511 = P.Mul()(opt_split_1509_1, opt_split_1496_1)
+ opt_add_1512 = opt_mul_1510 + self.add_1512_bias
+ opt_add_1513 = P.Add()(opt_add_1512, opt_mul_1511)
+ opt_conv2d_1514 = self.conv2d_1514(opt_add_1513)
+ opt_add_1515 = P.Add()(opt_conv2d_1514, opt_relu_1491)
+ opt_relu_1516 = self.relu_1516(opt_add_1515)
+ module22_61_opt = self.module22_61(opt_relu_1516)
+ opt_split_1521, opt_split_1521_1 = self.split_1521(module22_61_opt)
+ opt_add_1522 = opt_split_1521 + self.add_1522_bias
+ opt_add_1523 = P.Add()(opt_add_1522, opt_split_1521_1)
+ module20_60_opt = self.module20_60(opt_add_1523)
+ opt_reshape_1533 = self.reshape_1533(module20_60_opt, self.reshape_1533_shape)
+ opt_split_1534, opt_split_1534_1 = self.split_1534(opt_reshape_1533)
+ opt_mul_1535 = P.Mul()(opt_split_1534, opt_split_1521)
+ opt_mul_1536 = P.Mul()(opt_split_1534_1, opt_split_1521_1)
+ opt_add_1537 = opt_mul_1535 + self.add_1537_bias
+ opt_add_1538 = P.Add()(opt_add_1537, opt_mul_1536)
+ opt_conv2d_1539 = self.conv2d_1539(opt_add_1538)
+ opt_add_1540 = P.Add()(opt_conv2d_1539, opt_relu_1516)
+ opt_relu_1541 = self.relu_1541(opt_add_1540)
+ module22_62_opt = self.module22_62(opt_relu_1541)
+ opt_split_1546, opt_split_1546_1 = self.split_1546(module22_62_opt)
+ opt_add_1547 = opt_split_1546 + self.add_1547_bias
+ opt_add_1548 = P.Add()(opt_add_1547, opt_split_1546_1)
+ module20_61_opt = self.module20_61(opt_add_1548)
+ opt_reshape_1558 = self.reshape_1558(module20_61_opt, self.reshape_1558_shape)
+ opt_split_1559, opt_split_1559_1 = self.split_1559(opt_reshape_1558)
+ opt_mul_1560 = P.Mul()(opt_split_1559, opt_split_1546)
+ opt_mul_1561 = P.Mul()(opt_split_1559_1, opt_split_1546_1)
+ opt_add_1562 = opt_mul_1560 + self.add_1562_bias
+ opt_add_1563 = P.Add()(opt_add_1562, opt_mul_1561)
+ opt_conv2d_1564 = self.conv2d_1564(opt_add_1563)
+ opt_add_1565 = P.Add()(opt_conv2d_1564, opt_relu_1541)
+ opt_relu_1566 = self.relu_1566(opt_add_1565)
+ module22_63_opt = self.module22_63(opt_relu_1566)
+ opt_split_1571, opt_split_1571_1 = self.split_1571(module22_63_opt)
+ opt_add_1572 = opt_split_1571 + self.add_1572_bias
+ opt_add_1573 = P.Add()(opt_add_1572, opt_split_1571_1)
+ module20_62_opt = self.module20_62(opt_add_1573)
+ opt_reshape_1583 = self.reshape_1583(module20_62_opt, self.reshape_1583_shape)
+ opt_split_1584, opt_split_1584_1 = self.split_1584(opt_reshape_1583)
+ opt_mul_1585 = P.Mul()(opt_split_1584, opt_split_1571)
+ opt_mul_1586 = P.Mul()(opt_split_1584_1, opt_split_1571_1)
+ opt_add_1587 = opt_mul_1585 + self.add_1587_bias
+ opt_add_1588 = P.Add()(opt_add_1587, opt_mul_1586)
+ opt_conv2d_1589 = self.conv2d_1589(opt_add_1588)
+ opt_add_1590 = P.Add()(opt_conv2d_1589, opt_relu_1566)
+ opt_relu_1591 = self.relu_1591(opt_add_1590)
+ module22_64_opt = self.module22_64(opt_relu_1591)
+ opt_split_1598, opt_split_1598_1 = self.split_1598(module22_64_opt)
+ opt_add_1599 = opt_split_1598 + self.add_1599_bias
+ opt_add_1600 = P.Add()(opt_add_1599, opt_split_1598_1)
+ module20_63_opt = self.module20_63(opt_add_1600)
+ opt_reshape_1610 = self.reshape_1610(module20_63_opt, self.reshape_1610_shape)
+ opt_split_1611, opt_split_1611_1 = self.split_1611(opt_reshape_1610)
+ opt_mul_1612 = P.Mul()(opt_split_1611, opt_split_1598)
+ opt_mul_1613 = P.Mul()(opt_split_1611_1, opt_split_1598_1)
+ opt_add_1614 = opt_mul_1612 + self.add_1614_bias
+ opt_add_1615 = P.Add()(opt_add_1614, opt_mul_1613)
+ opt_pad_1616 = self.pad_1616(opt_add_1615)
+ opt_avgpool2d_1617 = self.pad_avgpool2d_1617(opt_pad_1616)
+ opt_avgpool2d_1617 = self.avgpool2d_1617(opt_avgpool2d_1617)
+ opt_conv2d_1618 = self.conv2d_1618(opt_avgpool2d_1617)
+ opt_avgpool2d_1593 = self.pad_avgpool2d_1593(opt_relu_1591)
+ opt_avgpool2d_1593 = self.avgpool2d_1593(opt_avgpool2d_1593)
+ opt_conv2d_1595 = self.conv2d_1595(opt_avgpool2d_1593)
+ opt_add_1619 = P.Add()(opt_conv2d_1618, opt_conv2d_1595)
+ opt_relu_1620 = self.relu_1620(opt_add_1619)
+ module22_65_opt = self.module22_65(opt_relu_1620)
+ opt_split_1625, opt_split_1625_1 = self.split_1625(module22_65_opt)
+ opt_add_1626 = opt_split_1625 + self.add_1626_bias
+ opt_add_1627 = P.Add()(opt_add_1626, opt_split_1625_1)
+ module20_64_opt = self.module20_64(opt_add_1627)
+ opt_reshape_1637 = self.reshape_1637(module20_64_opt, self.reshape_1637_shape)
+ opt_split_1638, opt_split_1638_1 = self.split_1638(opt_reshape_1637)
+ opt_mul_1639 = P.Mul()(opt_split_1638, opt_split_1625)
+ opt_mul_1640 = P.Mul()(opt_split_1638_1, opt_split_1625_1)
+ opt_add_1641 = opt_mul_1639 + self.add_1641_bias
+ opt_add_1642 = P.Add()(opt_add_1641, opt_mul_1640)
+ opt_conv2d_1643 = self.conv2d_1643(opt_add_1642)
+ opt_add_1644 = P.Add()(opt_conv2d_1643, opt_relu_1620)
+ opt_relu_1645 = self.relu_1645(opt_add_1644)
+ module22_66_opt = self.module22_66(opt_relu_1645)
+ opt_split_1650, opt_split_1650_1 = self.split_1650(module22_66_opt)
+ opt_add_1651 = opt_split_1650 + self.add_1651_bias
+ opt_add_1652 = P.Add()(opt_add_1651, opt_split_1650_1)
+ module20_65_opt = self.module20_65(opt_add_1652)
+ opt_reshape_1662 = self.reshape_1662(module20_65_opt, self.reshape_1662_shape)
+ opt_split_1663, opt_split_1663_1 = self.split_1663(opt_reshape_1662)
+ opt_mul_1664 = P.Mul()(opt_split_1663, opt_split_1650)
+ opt_mul_1665 = P.Mul()(opt_split_1663_1, opt_split_1650_1)
+ opt_add_1666 = opt_mul_1664 + self.add_1666_bias
+ opt_add_1667 = P.Add()(opt_add_1666, opt_mul_1665)
+ opt_conv2d_1668 = self.conv2d_1668(opt_add_1667)
+ opt_add_1669 = P.Add()(opt_conv2d_1668, opt_relu_1645)
+ opt_relu_1670 = self.relu_1670(opt_add_1669)
+ opt_avgpool2d_1671 = self.avgpool2d_1671(opt_relu_1670)
+ opt_reshape_1672 = self.reshape_1672(opt_avgpool2d_1671, self.reshape_1672_shape)
+ opt_flatten_1673 = self.flatten_1673(opt_reshape_1672)
+ opt_dense_1674 = self.dense_1674(opt_flatten_1673)
+ return opt_dense_1674
diff --git a/research/cvtmodel/resnest/src/resnest269.py b/research/cvtmodel/resnest/src/resnest269.py
new file mode 100644
index 0000000000000000000000000000000000000000..977564fc2a2764bb0f7d4137b3d400c17f9aa0b7
--- /dev/null
+++ b/research/cvtmodel/resnest/src/resnest269.py
@@ -0,0 +1,4591 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module5(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_0_kernel_size, conv2d_0_padding,
+ conv2d_0_pad_mode, conv2d_0_group):
+ super(Module5, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=conv2d_0_kernel_size,
+ stride=(1, 1),
+ padding=conv2d_0_padding,
+ pad_mode=conv2d_0_pad_mode,
+ dilation=(1, 1),
+ group=conv2d_0_group,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ return opt_relu_1
+
+
+class Module22(nn.Cell):
+ def __init__(self, module5_0_conv2d_0_in_channels, module5_0_conv2d_0_out_channels, module5_0_conv2d_0_kernel_size,
+ module5_0_conv2d_0_padding, module5_0_conv2d_0_pad_mode, module5_0_conv2d_0_group,
+ module5_1_conv2d_0_in_channels, module5_1_conv2d_0_out_channels, module5_1_conv2d_0_kernel_size,
+ module5_1_conv2d_0_padding, module5_1_conv2d_0_pad_mode, module5_1_conv2d_0_group):
+ super(Module22, self).__init__()
+ self.module5_0 = Module5(conv2d_0_in_channels=module5_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_0_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_0_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_0_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_0_conv2d_0_pad_mode,
+ conv2d_0_group=module5_0_conv2d_0_group)
+ self.module5_1 = Module5(conv2d_0_in_channels=module5_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_1_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_1_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_1_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_1_conv2d_0_pad_mode,
+ conv2d_0_group=module5_1_conv2d_0_group)
+
+ def construct(self, x):
+ module5_0_opt = self.module5_0(x)
+ module5_1_opt = self.module5_1(module5_0_opt)
+ return module5_1_opt
+
+
+class Module0(nn.Cell):
+ def __init__(self, avgpool2d_0_kernel_size, conv2d_1_in_channels, conv2d_1_out_channels, conv2d_3_in_channels,
+ conv2d_3_out_channels, reshape_4_shape):
+ super(Module0, self).__init__()
+ self.avgpool2d_0 = nn.AvgPool2d(kernel_size=avgpool2d_0_kernel_size)
+ self.conv2d_1 = nn.Conv2d(in_channels=conv2d_1_in_channels,
+ out_channels=conv2d_1_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=conv2d_3_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.reshape_4 = P.Reshape()
+ self.reshape_4_shape = tuple(reshape_4_shape)
+ self.transpose_5 = P.Transpose()
+
+ def construct(self, x):
+ opt_avgpool2d_0 = self.avgpool2d_0(x)
+ opt_conv2d_1 = self.conv2d_1(opt_avgpool2d_0)
+ opt_relu_2 = self.relu_2(opt_conv2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_reshape_4 = self.reshape_4(opt_conv2d_3, self.reshape_4_shape)
+ opt_transpose_5 = self.transpose_5(opt_reshape_4, (0, 3, 1, 2))
+ return opt_transpose_5
+
+
+class Module20(nn.Cell):
+ def __init__(self, reshape_2_shape, module0_0_avgpool2d_0_kernel_size, module0_0_conv2d_1_in_channels,
+ module0_0_conv2d_1_out_channels, module0_0_conv2d_3_in_channels, module0_0_conv2d_3_out_channels,
+ module0_0_reshape_4_shape):
+ super(Module20, self).__init__()
+ self.module0_0 = Module0(avgpool2d_0_kernel_size=module0_0_avgpool2d_0_kernel_size,
+ conv2d_1_in_channels=module0_0_conv2d_1_in_channels,
+ conv2d_1_out_channels=module0_0_conv2d_1_out_channels,
+ conv2d_3_in_channels=module0_0_conv2d_3_in_channels,
+ conv2d_3_out_channels=module0_0_conv2d_3_out_channels,
+ reshape_4_shape=module0_0_reshape_4_shape)
+ self.softmax_0 = nn.Softmax(axis=3)
+ self.transpose_1 = P.Transpose()
+ self.reshape_2 = P.Reshape()
+ self.reshape_2_shape = tuple(reshape_2_shape)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ opt_softmax_0 = self.softmax_0(module0_0_opt)
+ opt_transpose_1 = self.transpose_1(opt_softmax_0, (0, 3, 2, 1))
+ opt_reshape_2 = self.reshape_2(opt_transpose_1, self.reshape_2_shape)
+ return opt_reshape_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(3, 3),
+ stride=(2, 2),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.module22_0 = Module22(module5_0_conv2d_0_in_channels=64,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(3, 3),
+ module5_0_conv2d_0_padding=(1, 1, 1, 1),
+ module5_0_conv2d_0_pad_mode="pad",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=1)
+ self.pad_maxpool2d_6 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_6 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module22_1 = Module22(module5_0_conv2d_0_in_channels=128,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_13 = P.Split(axis=1, output_num=2)
+ self.add_14_bias = 0.0
+ self.module20_0 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(104, 104),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_25 = P.Reshape()
+ self.reshape_25_shape = tuple([1, 128, 1, 1])
+ self.split_26 = P.Split(axis=1, output_num=2)
+ self.add_29_bias = 0.0
+ self.conv2d_31 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_8 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_8 = nn.AvgPool2d(kernel_size=(1, 1), stride=(1, 1))
+ self.conv2d_10 = nn.Conv2d(in_channels=128,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_33 = nn.ReLU()
+ self.module22_2 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_38 = P.Split(axis=1, output_num=2)
+ self.add_39_bias = 0.0
+ self.module20_1 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(104, 104),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_50 = P.Reshape()
+ self.reshape_50_shape = tuple([1, 128, 1, 1])
+ self.split_51 = P.Split(axis=1, output_num=2)
+ self.add_54_bias = 0.0
+ self.conv2d_56 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_58 = nn.ReLU()
+ self.module22_3 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_63 = P.Split(axis=1, output_num=2)
+ self.add_64_bias = 0.0
+ self.module20_2 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(104, 104),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_75 = P.Reshape()
+ self.reshape_75_shape = tuple([1, 128, 1, 1])
+ self.split_76 = P.Split(axis=1, output_num=2)
+ self.add_79_bias = 0.0
+ self.conv2d_81 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_83 = nn.ReLU()
+ self.module22_4 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_90 = P.Split(axis=1, output_num=2)
+ self.add_91_bias = 0.0
+ self.module20_3 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(104, 104),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_102 = P.Reshape()
+ self.reshape_102_shape = tuple([1, 256, 1, 1])
+ self.split_103 = P.Split(axis=1, output_num=2)
+ self.add_106_bias = 0.0
+ self.pad_108 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_109 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_109 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_110 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_85 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_85 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_87 = nn.Conv2d(in_channels=256,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_112 = nn.ReLU()
+ self.module22_5 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_117 = P.Split(axis=1, output_num=2)
+ self.add_118_bias = 0.0
+ self.module20_4 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_129 = P.Reshape()
+ self.reshape_129_shape = tuple([1, 256, 1, 1])
+ self.split_130 = P.Split(axis=1, output_num=2)
+ self.add_133_bias = 0.0
+ self.conv2d_135 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_137 = nn.ReLU()
+ self.module22_6 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_142 = P.Split(axis=1, output_num=2)
+ self.add_143_bias = 0.0
+ self.module20_5 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_154 = P.Reshape()
+ self.reshape_154_shape = tuple([1, 256, 1, 1])
+ self.split_155 = P.Split(axis=1, output_num=2)
+ self.add_158_bias = 0.0
+ self.conv2d_160 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_162 = nn.ReLU()
+ self.module22_7 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_167 = P.Split(axis=1, output_num=2)
+ self.add_168_bias = 0.0
+ self.module20_6 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_179 = P.Reshape()
+ self.reshape_179_shape = tuple([1, 256, 1, 1])
+ self.split_180 = P.Split(axis=1, output_num=2)
+ self.add_183_bias = 0.0
+ self.conv2d_185 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_187 = nn.ReLU()
+ self.module22_8 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_192 = P.Split(axis=1, output_num=2)
+ self.add_193_bias = 0.0
+ self.module20_7 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_204 = P.Reshape()
+ self.reshape_204_shape = tuple([1, 256, 1, 1])
+ self.split_205 = P.Split(axis=1, output_num=2)
+ self.add_208_bias = 0.0
+ self.conv2d_210 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_212 = nn.ReLU()
+ self.module22_9 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_217 = P.Split(axis=1, output_num=2)
+ self.add_218_bias = 0.0
+ self.module20_8 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_229 = P.Reshape()
+ self.reshape_229_shape = tuple([1, 256, 1, 1])
+ self.split_230 = P.Split(axis=1, output_num=2)
+ self.add_233_bias = 0.0
+ self.conv2d_235 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_237 = nn.ReLU()
+ self.module22_10 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_242 = P.Split(axis=1, output_num=2)
+ self.add_243_bias = 0.0
+ self.module20_9 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_254 = P.Reshape()
+ self.reshape_254_shape = tuple([1, 256, 1, 1])
+ self.split_255 = P.Split(axis=1, output_num=2)
+ self.add_258_bias = 0.0
+ self.conv2d_260 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_262 = nn.ReLU()
+ self.module22_11 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_267 = P.Split(axis=1, output_num=2)
+ self.add_268_bias = 0.0
+ self.module20_10 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_279 = P.Reshape()
+ self.reshape_279_shape = tuple([1, 256, 1, 1])
+ self.split_280 = P.Split(axis=1, output_num=2)
+ self.add_283_bias = 0.0
+ self.conv2d_285 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_287 = nn.ReLU()
+ self.module22_12 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_292 = P.Split(axis=1, output_num=2)
+ self.add_293_bias = 0.0
+ self.module20_11 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_304 = P.Reshape()
+ self.reshape_304_shape = tuple([1, 256, 1, 1])
+ self.split_305 = P.Split(axis=1, output_num=2)
+ self.add_308_bias = 0.0
+ self.conv2d_310 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_312 = nn.ReLU()
+ self.module22_13 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_317 = P.Split(axis=1, output_num=2)
+ self.add_318_bias = 0.0
+ self.module20_12 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_329 = P.Reshape()
+ self.reshape_329_shape = tuple([1, 256, 1, 1])
+ self.split_330 = P.Split(axis=1, output_num=2)
+ self.add_333_bias = 0.0
+ self.conv2d_335 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_337 = nn.ReLU()
+ self.module22_14 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_342 = P.Split(axis=1, output_num=2)
+ self.add_343_bias = 0.0
+ self.module20_13 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_354 = P.Reshape()
+ self.reshape_354_shape = tuple([1, 256, 1, 1])
+ self.split_355 = P.Split(axis=1, output_num=2)
+ self.add_358_bias = 0.0
+ self.conv2d_360 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_362 = nn.ReLU()
+ self.module22_15 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_367 = P.Split(axis=1, output_num=2)
+ self.add_368_bias = 0.0
+ self.module20_14 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_379 = P.Reshape()
+ self.reshape_379_shape = tuple([1, 256, 1, 1])
+ self.split_380 = P.Split(axis=1, output_num=2)
+ self.add_383_bias = 0.0
+ self.conv2d_385 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_387 = nn.ReLU()
+ self.module22_16 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_392 = P.Split(axis=1, output_num=2)
+ self.add_393_bias = 0.0
+ self.module20_15 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_404 = P.Reshape()
+ self.reshape_404_shape = tuple([1, 256, 1, 1])
+ self.split_405 = P.Split(axis=1, output_num=2)
+ self.add_408_bias = 0.0
+ self.conv2d_410 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_412 = nn.ReLU()
+ self.module22_17 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_417 = P.Split(axis=1, output_num=2)
+ self.add_418_bias = 0.0
+ self.module20_16 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_429 = P.Reshape()
+ self.reshape_429_shape = tuple([1, 256, 1, 1])
+ self.split_430 = P.Split(axis=1, output_num=2)
+ self.add_433_bias = 0.0
+ self.conv2d_435 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_437 = nn.ReLU()
+ self.module22_18 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_442 = P.Split(axis=1, output_num=2)
+ self.add_443_bias = 0.0
+ self.module20_17 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_454 = P.Reshape()
+ self.reshape_454_shape = tuple([1, 256, 1, 1])
+ self.split_455 = P.Split(axis=1, output_num=2)
+ self.add_458_bias = 0.0
+ self.conv2d_460 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_462 = nn.ReLU()
+ self.module22_19 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_467 = P.Split(axis=1, output_num=2)
+ self.add_468_bias = 0.0
+ self.module20_18 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_479 = P.Reshape()
+ self.reshape_479_shape = tuple([1, 256, 1, 1])
+ self.split_480 = P.Split(axis=1, output_num=2)
+ self.add_483_bias = 0.0
+ self.conv2d_485 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_487 = nn.ReLU()
+ self.module22_20 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_492 = P.Split(axis=1, output_num=2)
+ self.add_493_bias = 0.0
+ self.module20_19 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_504 = P.Reshape()
+ self.reshape_504_shape = tuple([1, 256, 1, 1])
+ self.split_505 = P.Split(axis=1, output_num=2)
+ self.add_508_bias = 0.0
+ self.conv2d_510 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_512 = nn.ReLU()
+ self.module22_21 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_517 = P.Split(axis=1, output_num=2)
+ self.add_518_bias = 0.0
+ self.module20_20 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_529 = P.Reshape()
+ self.reshape_529_shape = tuple([1, 256, 1, 1])
+ self.split_530 = P.Split(axis=1, output_num=2)
+ self.add_533_bias = 0.0
+ self.conv2d_535 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_537 = nn.ReLU()
+ self.module22_22 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_542 = P.Split(axis=1, output_num=2)
+ self.add_543_bias = 0.0
+ self.module20_21 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_554 = P.Reshape()
+ self.reshape_554_shape = tuple([1, 256, 1, 1])
+ self.split_555 = P.Split(axis=1, output_num=2)
+ self.add_558_bias = 0.0
+ self.conv2d_560 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_562 = nn.ReLU()
+ self.module22_23 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_567 = P.Split(axis=1, output_num=2)
+ self.add_568_bias = 0.0
+ self.module20_22 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_579 = P.Reshape()
+ self.reshape_579_shape = tuple([1, 256, 1, 1])
+ self.split_580 = P.Split(axis=1, output_num=2)
+ self.add_583_bias = 0.0
+ self.conv2d_585 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_587 = nn.ReLU()
+ self.module22_24 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_592 = P.Split(axis=1, output_num=2)
+ self.add_593_bias = 0.0
+ self.module20_23 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_604 = P.Reshape()
+ self.reshape_604_shape = tuple([1, 256, 1, 1])
+ self.split_605 = P.Split(axis=1, output_num=2)
+ self.add_608_bias = 0.0
+ self.conv2d_610 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_612 = nn.ReLU()
+ self.module22_25 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_617 = P.Split(axis=1, output_num=2)
+ self.add_618_bias = 0.0
+ self.module20_24 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_629 = P.Reshape()
+ self.reshape_629_shape = tuple([1, 256, 1, 1])
+ self.split_630 = P.Split(axis=1, output_num=2)
+ self.add_633_bias = 0.0
+ self.conv2d_635 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_637 = nn.ReLU()
+ self.module22_26 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_642 = P.Split(axis=1, output_num=2)
+ self.add_643_bias = 0.0
+ self.module20_25 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_654 = P.Reshape()
+ self.reshape_654_shape = tuple([1, 256, 1, 1])
+ self.split_655 = P.Split(axis=1, output_num=2)
+ self.add_658_bias = 0.0
+ self.conv2d_660 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_662 = nn.ReLU()
+ self.module22_27 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_667 = P.Split(axis=1, output_num=2)
+ self.add_668_bias = 0.0
+ self.module20_26 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_679 = P.Reshape()
+ self.reshape_679_shape = tuple([1, 256, 1, 1])
+ self.split_680 = P.Split(axis=1, output_num=2)
+ self.add_683_bias = 0.0
+ self.conv2d_685 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_687 = nn.ReLU()
+ self.module22_28 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_692 = P.Split(axis=1, output_num=2)
+ self.add_693_bias = 0.0
+ self.module20_27 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_704 = P.Reshape()
+ self.reshape_704_shape = tuple([1, 256, 1, 1])
+ self.split_705 = P.Split(axis=1, output_num=2)
+ self.add_708_bias = 0.0
+ self.conv2d_710 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_712 = nn.ReLU()
+ self.module22_29 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_717 = P.Split(axis=1, output_num=2)
+ self.add_718_bias = 0.0
+ self.module20_28 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_729 = P.Reshape()
+ self.reshape_729_shape = tuple([1, 256, 1, 1])
+ self.split_730 = P.Split(axis=1, output_num=2)
+ self.add_733_bias = 0.0
+ self.conv2d_735 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_737 = nn.ReLU()
+ self.module22_30 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_742 = P.Split(axis=1, output_num=2)
+ self.add_743_bias = 0.0
+ self.module20_29 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_754 = P.Reshape()
+ self.reshape_754_shape = tuple([1, 256, 1, 1])
+ self.split_755 = P.Split(axis=1, output_num=2)
+ self.add_758_bias = 0.0
+ self.conv2d_760 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_762 = nn.ReLU()
+ self.module22_31 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_767 = P.Split(axis=1, output_num=2)
+ self.add_768_bias = 0.0
+ self.module20_30 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_779 = P.Reshape()
+ self.reshape_779_shape = tuple([1, 256, 1, 1])
+ self.split_780 = P.Split(axis=1, output_num=2)
+ self.add_783_bias = 0.0
+ self.conv2d_785 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_787 = nn.ReLU()
+ self.module22_32 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_792 = P.Split(axis=1, output_num=2)
+ self.add_793_bias = 0.0
+ self.module20_31 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_804 = P.Reshape()
+ self.reshape_804_shape = tuple([1, 256, 1, 1])
+ self.split_805 = P.Split(axis=1, output_num=2)
+ self.add_808_bias = 0.0
+ self.conv2d_810 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_812 = nn.ReLU()
+ self.module22_33 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_817 = P.Split(axis=1, output_num=2)
+ self.add_818_bias = 0.0
+ self.module20_32 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_829 = P.Reshape()
+ self.reshape_829_shape = tuple([1, 256, 1, 1])
+ self.split_830 = P.Split(axis=1, output_num=2)
+ self.add_833_bias = 0.0
+ self.conv2d_835 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_837 = nn.ReLU()
+ self.module22_34 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_844 = P.Split(axis=1, output_num=2)
+ self.add_845_bias = 0.0
+ self.module20_33 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(52, 52),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_856 = P.Reshape()
+ self.reshape_856_shape = tuple([1, 512, 1, 1])
+ self.split_857 = P.Split(axis=1, output_num=2)
+ self.add_860_bias = 0.0
+ self.pad_862 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_863 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_863 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_864 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_839 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_839 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_841 = nn.Conv2d(in_channels=512,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_866 = nn.ReLU()
+ self.module22_35 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_871 = P.Split(axis=1, output_num=2)
+ self.add_872_bias = 0.0
+ self.module20_34 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_883 = P.Reshape()
+ self.reshape_883_shape = tuple([1, 512, 1, 1])
+ self.split_884 = P.Split(axis=1, output_num=2)
+ self.add_887_bias = 0.0
+ self.conv2d_889 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_891 = nn.ReLU()
+ self.module22_36 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_896 = P.Split(axis=1, output_num=2)
+ self.add_897_bias = 0.0
+ self.module20_35 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_908 = P.Reshape()
+ self.reshape_908_shape = tuple([1, 512, 1, 1])
+ self.split_909 = P.Split(axis=1, output_num=2)
+ self.add_912_bias = 0.0
+ self.conv2d_914 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_916 = nn.ReLU()
+ self.module22_37 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_921 = P.Split(axis=1, output_num=2)
+ self.add_922_bias = 0.0
+ self.module20_36 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_933 = P.Reshape()
+ self.reshape_933_shape = tuple([1, 512, 1, 1])
+ self.split_934 = P.Split(axis=1, output_num=2)
+ self.add_937_bias = 0.0
+ self.conv2d_939 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_941 = nn.ReLU()
+ self.module22_38 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_946 = P.Split(axis=1, output_num=2)
+ self.add_947_bias = 0.0
+ self.module20_37 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_958 = P.Reshape()
+ self.reshape_958_shape = tuple([1, 512, 1, 1])
+ self.split_959 = P.Split(axis=1, output_num=2)
+ self.add_962_bias = 0.0
+ self.conv2d_964 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_966 = nn.ReLU()
+ self.module22_39 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_971 = P.Split(axis=1, output_num=2)
+ self.add_972_bias = 0.0
+ self.module20_38 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_983 = P.Reshape()
+ self.reshape_983_shape = tuple([1, 512, 1, 1])
+ self.split_984 = P.Split(axis=1, output_num=2)
+ self.add_987_bias = 0.0
+ self.conv2d_989 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_991 = nn.ReLU()
+ self.module22_40 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_996 = P.Split(axis=1, output_num=2)
+ self.add_997_bias = 0.0
+ self.module20_39 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1008 = P.Reshape()
+ self.reshape_1008_shape = tuple([1, 512, 1, 1])
+ self.split_1009 = P.Split(axis=1, output_num=2)
+ self.add_1012_bias = 0.0
+ self.conv2d_1014 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1016 = nn.ReLU()
+ self.module22_41 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1021 = P.Split(axis=1, output_num=2)
+ self.add_1022_bias = 0.0
+ self.module20_40 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1033 = P.Reshape()
+ self.reshape_1033_shape = tuple([1, 512, 1, 1])
+ self.split_1034 = P.Split(axis=1, output_num=2)
+ self.add_1037_bias = 0.0
+ self.conv2d_1039 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1041 = nn.ReLU()
+ self.module22_42 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1046 = P.Split(axis=1, output_num=2)
+ self.add_1047_bias = 0.0
+ self.module20_41 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1058 = P.Reshape()
+ self.reshape_1058_shape = tuple([1, 512, 1, 1])
+ self.split_1059 = P.Split(axis=1, output_num=2)
+ self.add_1062_bias = 0.0
+ self.conv2d_1064 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1066 = nn.ReLU()
+ self.module22_43 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1071 = P.Split(axis=1, output_num=2)
+ self.add_1072_bias = 0.0
+ self.module20_42 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1083 = P.Reshape()
+ self.reshape_1083_shape = tuple([1, 512, 1, 1])
+ self.split_1084 = P.Split(axis=1, output_num=2)
+ self.add_1087_bias = 0.0
+ self.conv2d_1089 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1091 = nn.ReLU()
+ self.module22_44 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1096 = P.Split(axis=1, output_num=2)
+ self.add_1097_bias = 0.0
+ self.module20_43 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1108 = P.Reshape()
+ self.reshape_1108_shape = tuple([1, 512, 1, 1])
+ self.split_1109 = P.Split(axis=1, output_num=2)
+ self.add_1112_bias = 0.0
+ self.conv2d_1114 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1116 = nn.ReLU()
+ self.module22_45 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1121 = P.Split(axis=1, output_num=2)
+ self.add_1122_bias = 0.0
+ self.module20_44 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1133 = P.Reshape()
+ self.reshape_1133_shape = tuple([1, 512, 1, 1])
+ self.split_1134 = P.Split(axis=1, output_num=2)
+ self.add_1137_bias = 0.0
+ self.conv2d_1139 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1141 = nn.ReLU()
+ self.module22_46 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1146 = P.Split(axis=1, output_num=2)
+ self.add_1147_bias = 0.0
+ self.module20_45 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1158 = P.Reshape()
+ self.reshape_1158_shape = tuple([1, 512, 1, 1])
+ self.split_1159 = P.Split(axis=1, output_num=2)
+ self.add_1162_bias = 0.0
+ self.conv2d_1164 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1166 = nn.ReLU()
+ self.module22_47 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1171 = P.Split(axis=1, output_num=2)
+ self.add_1172_bias = 0.0
+ self.module20_46 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1183 = P.Reshape()
+ self.reshape_1183_shape = tuple([1, 512, 1, 1])
+ self.split_1184 = P.Split(axis=1, output_num=2)
+ self.add_1187_bias = 0.0
+ self.conv2d_1189 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1191 = nn.ReLU()
+ self.module22_48 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1196 = P.Split(axis=1, output_num=2)
+ self.add_1197_bias = 0.0
+ self.module20_47 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1208 = P.Reshape()
+ self.reshape_1208_shape = tuple([1, 512, 1, 1])
+ self.split_1209 = P.Split(axis=1, output_num=2)
+ self.add_1212_bias = 0.0
+ self.conv2d_1214 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1216 = nn.ReLU()
+ self.module22_49 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1221 = P.Split(axis=1, output_num=2)
+ self.add_1222_bias = 0.0
+ self.module20_48 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1233 = P.Reshape()
+ self.reshape_1233_shape = tuple([1, 512, 1, 1])
+ self.split_1234 = P.Split(axis=1, output_num=2)
+ self.add_1237_bias = 0.0
+ self.conv2d_1239 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1241 = nn.ReLU()
+ self.module22_50 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1246 = P.Split(axis=1, output_num=2)
+ self.add_1247_bias = 0.0
+ self.module20_49 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1258 = P.Reshape()
+ self.reshape_1258_shape = tuple([1, 512, 1, 1])
+ self.split_1259 = P.Split(axis=1, output_num=2)
+ self.add_1262_bias = 0.0
+ self.conv2d_1264 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1266 = nn.ReLU()
+ self.module22_51 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1271 = P.Split(axis=1, output_num=2)
+ self.add_1272_bias = 0.0
+ self.module20_50 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1283 = P.Reshape()
+ self.reshape_1283_shape = tuple([1, 512, 1, 1])
+ self.split_1284 = P.Split(axis=1, output_num=2)
+ self.add_1287_bias = 0.0
+ self.conv2d_1289 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1291 = nn.ReLU()
+ self.module22_52 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1296 = P.Split(axis=1, output_num=2)
+ self.add_1297_bias = 0.0
+ self.module20_51 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1308 = P.Reshape()
+ self.reshape_1308_shape = tuple([1, 512, 1, 1])
+ self.split_1309 = P.Split(axis=1, output_num=2)
+ self.add_1312_bias = 0.0
+ self.conv2d_1314 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1316 = nn.ReLU()
+ self.module22_53 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1321 = P.Split(axis=1, output_num=2)
+ self.add_1322_bias = 0.0
+ self.module20_52 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1333 = P.Reshape()
+ self.reshape_1333_shape = tuple([1, 512, 1, 1])
+ self.split_1334 = P.Split(axis=1, output_num=2)
+ self.add_1337_bias = 0.0
+ self.conv2d_1339 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1341 = nn.ReLU()
+ self.module22_54 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1346 = P.Split(axis=1, output_num=2)
+ self.add_1347_bias = 0.0
+ self.module20_53 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1358 = P.Reshape()
+ self.reshape_1358_shape = tuple([1, 512, 1, 1])
+ self.split_1359 = P.Split(axis=1, output_num=2)
+ self.add_1362_bias = 0.0
+ self.conv2d_1364 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1366 = nn.ReLU()
+ self.module22_55 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1371 = P.Split(axis=1, output_num=2)
+ self.add_1372_bias = 0.0
+ self.module20_54 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1383 = P.Reshape()
+ self.reshape_1383_shape = tuple([1, 512, 1, 1])
+ self.split_1384 = P.Split(axis=1, output_num=2)
+ self.add_1387_bias = 0.0
+ self.conv2d_1389 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1391 = nn.ReLU()
+ self.module22_56 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1396 = P.Split(axis=1, output_num=2)
+ self.add_1397_bias = 0.0
+ self.module20_55 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1408 = P.Reshape()
+ self.reshape_1408_shape = tuple([1, 512, 1, 1])
+ self.split_1409 = P.Split(axis=1, output_num=2)
+ self.add_1412_bias = 0.0
+ self.conv2d_1414 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1416 = nn.ReLU()
+ self.module22_57 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1421 = P.Split(axis=1, output_num=2)
+ self.add_1422_bias = 0.0
+ self.module20_56 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1433 = P.Reshape()
+ self.reshape_1433_shape = tuple([1, 512, 1, 1])
+ self.split_1434 = P.Split(axis=1, output_num=2)
+ self.add_1437_bias = 0.0
+ self.conv2d_1439 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1441 = nn.ReLU()
+ self.module22_58 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1446 = P.Split(axis=1, output_num=2)
+ self.add_1447_bias = 0.0
+ self.module20_57 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1458 = P.Reshape()
+ self.reshape_1458_shape = tuple([1, 512, 1, 1])
+ self.split_1459 = P.Split(axis=1, output_num=2)
+ self.add_1462_bias = 0.0
+ self.conv2d_1464 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1466 = nn.ReLU()
+ self.module22_59 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1471 = P.Split(axis=1, output_num=2)
+ self.add_1472_bias = 0.0
+ self.module20_58 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1483 = P.Reshape()
+ self.reshape_1483_shape = tuple([1, 512, 1, 1])
+ self.split_1484 = P.Split(axis=1, output_num=2)
+ self.add_1487_bias = 0.0
+ self.conv2d_1489 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1491 = nn.ReLU()
+ self.module22_60 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1496 = P.Split(axis=1, output_num=2)
+ self.add_1497_bias = 0.0
+ self.module20_59 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1508 = P.Reshape()
+ self.reshape_1508_shape = tuple([1, 512, 1, 1])
+ self.split_1509 = P.Split(axis=1, output_num=2)
+ self.add_1512_bias = 0.0
+ self.conv2d_1514 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1516 = nn.ReLU()
+ self.module22_61 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1521 = P.Split(axis=1, output_num=2)
+ self.add_1522_bias = 0.0
+ self.module20_60 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1533 = P.Reshape()
+ self.reshape_1533_shape = tuple([1, 512, 1, 1])
+ self.split_1534 = P.Split(axis=1, output_num=2)
+ self.add_1537_bias = 0.0
+ self.conv2d_1539 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1541 = nn.ReLU()
+ self.module22_62 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1546 = P.Split(axis=1, output_num=2)
+ self.add_1547_bias = 0.0
+ self.module20_61 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1558 = P.Reshape()
+ self.reshape_1558_shape = tuple([1, 512, 1, 1])
+ self.split_1559 = P.Split(axis=1, output_num=2)
+ self.add_1562_bias = 0.0
+ self.conv2d_1564 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1566 = nn.ReLU()
+ self.module22_63 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1571 = P.Split(axis=1, output_num=2)
+ self.add_1572_bias = 0.0
+ self.module20_62 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1583 = P.Reshape()
+ self.reshape_1583_shape = tuple([1, 512, 1, 1])
+ self.split_1584 = P.Split(axis=1, output_num=2)
+ self.add_1587_bias = 0.0
+ self.conv2d_1589 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1591 = nn.ReLU()
+ self.module22_64 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1596 = P.Split(axis=1, output_num=2)
+ self.add_1597_bias = 0.0
+ self.module20_63 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1608 = P.Reshape()
+ self.reshape_1608_shape = tuple([1, 512, 1, 1])
+ self.split_1609 = P.Split(axis=1, output_num=2)
+ self.add_1612_bias = 0.0
+ self.conv2d_1614 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1616 = nn.ReLU()
+ self.module22_65 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1621 = P.Split(axis=1, output_num=2)
+ self.add_1622_bias = 0.0
+ self.module20_64 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1633 = P.Reshape()
+ self.reshape_1633_shape = tuple([1, 512, 1, 1])
+ self.split_1634 = P.Split(axis=1, output_num=2)
+ self.add_1637_bias = 0.0
+ self.conv2d_1639 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1641 = nn.ReLU()
+ self.module22_66 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1646 = P.Split(axis=1, output_num=2)
+ self.add_1647_bias = 0.0
+ self.module20_65 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1658 = P.Reshape()
+ self.reshape_1658_shape = tuple([1, 512, 1, 1])
+ self.split_1659 = P.Split(axis=1, output_num=2)
+ self.add_1662_bias = 0.0
+ self.conv2d_1664 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1666 = nn.ReLU()
+ self.module22_67 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1671 = P.Split(axis=1, output_num=2)
+ self.add_1672_bias = 0.0
+ self.module20_66 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1683 = P.Reshape()
+ self.reshape_1683_shape = tuple([1, 512, 1, 1])
+ self.split_1684 = P.Split(axis=1, output_num=2)
+ self.add_1687_bias = 0.0
+ self.conv2d_1689 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1691 = nn.ReLU()
+ self.module22_68 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1696 = P.Split(axis=1, output_num=2)
+ self.add_1697_bias = 0.0
+ self.module20_67 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1708 = P.Reshape()
+ self.reshape_1708_shape = tuple([1, 512, 1, 1])
+ self.split_1709 = P.Split(axis=1, output_num=2)
+ self.add_1712_bias = 0.0
+ self.conv2d_1714 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1716 = nn.ReLU()
+ self.module22_69 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1721 = P.Split(axis=1, output_num=2)
+ self.add_1722_bias = 0.0
+ self.module20_68 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1733 = P.Reshape()
+ self.reshape_1733_shape = tuple([1, 512, 1, 1])
+ self.split_1734 = P.Split(axis=1, output_num=2)
+ self.add_1737_bias = 0.0
+ self.conv2d_1739 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1741 = nn.ReLU()
+ self.module22_70 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1746 = P.Split(axis=1, output_num=2)
+ self.add_1747_bias = 0.0
+ self.module20_69 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1758 = P.Reshape()
+ self.reshape_1758_shape = tuple([1, 512, 1, 1])
+ self.split_1759 = P.Split(axis=1, output_num=2)
+ self.add_1762_bias = 0.0
+ self.conv2d_1764 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1766 = nn.ReLU()
+ self.module22_71 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1771 = P.Split(axis=1, output_num=2)
+ self.add_1772_bias = 0.0
+ self.module20_70 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1783 = P.Reshape()
+ self.reshape_1783_shape = tuple([1, 512, 1, 1])
+ self.split_1784 = P.Split(axis=1, output_num=2)
+ self.add_1787_bias = 0.0
+ self.conv2d_1789 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1791 = nn.ReLU()
+ self.module22_72 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1796 = P.Split(axis=1, output_num=2)
+ self.add_1797_bias = 0.0
+ self.module20_71 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1808 = P.Reshape()
+ self.reshape_1808_shape = tuple([1, 512, 1, 1])
+ self.split_1809 = P.Split(axis=1, output_num=2)
+ self.add_1812_bias = 0.0
+ self.conv2d_1814 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1816 = nn.ReLU()
+ self.module22_73 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1821 = P.Split(axis=1, output_num=2)
+ self.add_1822_bias = 0.0
+ self.module20_72 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1833 = P.Reshape()
+ self.reshape_1833_shape = tuple([1, 512, 1, 1])
+ self.split_1834 = P.Split(axis=1, output_num=2)
+ self.add_1837_bias = 0.0
+ self.conv2d_1839 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1841 = nn.ReLU()
+ self.module22_74 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1846 = P.Split(axis=1, output_num=2)
+ self.add_1847_bias = 0.0
+ self.module20_73 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1858 = P.Reshape()
+ self.reshape_1858_shape = tuple([1, 512, 1, 1])
+ self.split_1859 = P.Split(axis=1, output_num=2)
+ self.add_1862_bias = 0.0
+ self.conv2d_1864 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1866 = nn.ReLU()
+ self.module22_75 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1871 = P.Split(axis=1, output_num=2)
+ self.add_1872_bias = 0.0
+ self.module20_74 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1883 = P.Reshape()
+ self.reshape_1883_shape = tuple([1, 512, 1, 1])
+ self.split_1884 = P.Split(axis=1, output_num=2)
+ self.add_1887_bias = 0.0
+ self.conv2d_1889 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1891 = nn.ReLU()
+ self.module22_76 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1896 = P.Split(axis=1, output_num=2)
+ self.add_1897_bias = 0.0
+ self.module20_75 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1908 = P.Reshape()
+ self.reshape_1908_shape = tuple([1, 512, 1, 1])
+ self.split_1909 = P.Split(axis=1, output_num=2)
+ self.add_1912_bias = 0.0
+ self.conv2d_1914 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1916 = nn.ReLU()
+ self.module22_77 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1921 = P.Split(axis=1, output_num=2)
+ self.add_1922_bias = 0.0
+ self.module20_76 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1933 = P.Reshape()
+ self.reshape_1933_shape = tuple([1, 512, 1, 1])
+ self.split_1934 = P.Split(axis=1, output_num=2)
+ self.add_1937_bias = 0.0
+ self.conv2d_1939 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1941 = nn.ReLU()
+ self.module22_78 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1946 = P.Split(axis=1, output_num=2)
+ self.add_1947_bias = 0.0
+ self.module20_77 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1958 = P.Reshape()
+ self.reshape_1958_shape = tuple([1, 512, 1, 1])
+ self.split_1959 = P.Split(axis=1, output_num=2)
+ self.add_1962_bias = 0.0
+ self.conv2d_1964 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1966 = nn.ReLU()
+ self.module22_79 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1971 = P.Split(axis=1, output_num=2)
+ self.add_1972_bias = 0.0
+ self.module20_78 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_1983 = P.Reshape()
+ self.reshape_1983_shape = tuple([1, 512, 1, 1])
+ self.split_1984 = P.Split(axis=1, output_num=2)
+ self.add_1987_bias = 0.0
+ self.conv2d_1989 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1991 = nn.ReLU()
+ self.module22_80 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_1996 = P.Split(axis=1, output_num=2)
+ self.add_1997_bias = 0.0
+ self.module20_79 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_2008 = P.Reshape()
+ self.reshape_2008_shape = tuple([1, 512, 1, 1])
+ self.split_2009 = P.Split(axis=1, output_num=2)
+ self.add_2012_bias = 0.0
+ self.conv2d_2014 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2016 = nn.ReLU()
+ self.module22_81 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2021 = P.Split(axis=1, output_num=2)
+ self.add_2022_bias = 0.0
+ self.module20_80 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_2033 = P.Reshape()
+ self.reshape_2033_shape = tuple([1, 512, 1, 1])
+ self.split_2034 = P.Split(axis=1, output_num=2)
+ self.add_2037_bias = 0.0
+ self.conv2d_2039 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2041 = nn.ReLU()
+ self.module22_82 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2048 = P.Split(axis=1, output_num=2)
+ self.add_2049_bias = 0.0
+ self.module20_81 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(26, 26),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2060 = P.Reshape()
+ self.reshape_2060_shape = tuple([1, 1024, 1, 1])
+ self.split_2061 = P.Split(axis=1, output_num=2)
+ self.add_2064_bias = 0.0
+ self.pad_2066 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_2067 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_2067 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_2068 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_2043 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_2043 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_2045 = nn.Conv2d(in_channels=1024,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2070 = nn.ReLU()
+ self.module22_83 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2075 = P.Split(axis=1, output_num=2)
+ self.add_2076_bias = 0.0
+ self.module20_82 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2087 = P.Reshape()
+ self.reshape_2087_shape = tuple([1, 1024, 1, 1])
+ self.split_2088 = P.Split(axis=1, output_num=2)
+ self.add_2091_bias = 0.0
+ self.conv2d_2093 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2095 = nn.ReLU()
+ self.module22_84 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2100 = P.Split(axis=1, output_num=2)
+ self.add_2101_bias = 0.0
+ self.module20_83 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2112 = P.Reshape()
+ self.reshape_2112_shape = tuple([1, 1024, 1, 1])
+ self.split_2113 = P.Split(axis=1, output_num=2)
+ self.add_2116_bias = 0.0
+ self.conv2d_2118 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2120 = nn.ReLU()
+ self.module22_85 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2125 = P.Split(axis=1, output_num=2)
+ self.add_2126_bias = 0.0
+ self.module20_84 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2137 = P.Reshape()
+ self.reshape_2137_shape = tuple([1, 1024, 1, 1])
+ self.split_2138 = P.Split(axis=1, output_num=2)
+ self.add_2141_bias = 0.0
+ self.conv2d_2143 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2145 = nn.ReLU()
+ self.module22_86 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2150 = P.Split(axis=1, output_num=2)
+ self.add_2151_bias = 0.0
+ self.module20_85 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2162 = P.Reshape()
+ self.reshape_2162_shape = tuple([1, 1024, 1, 1])
+ self.split_2163 = P.Split(axis=1, output_num=2)
+ self.add_2166_bias = 0.0
+ self.conv2d_2168 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2170 = nn.ReLU()
+ self.module22_87 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2175 = P.Split(axis=1, output_num=2)
+ self.add_2176_bias = 0.0
+ self.module20_86 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2187 = P.Reshape()
+ self.reshape_2187_shape = tuple([1, 1024, 1, 1])
+ self.split_2188 = P.Split(axis=1, output_num=2)
+ self.add_2191_bias = 0.0
+ self.conv2d_2193 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2195 = nn.ReLU()
+ self.module22_88 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2200 = P.Split(axis=1, output_num=2)
+ self.add_2201_bias = 0.0
+ self.module20_87 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2212 = P.Reshape()
+ self.reshape_2212_shape = tuple([1, 1024, 1, 1])
+ self.split_2213 = P.Split(axis=1, output_num=2)
+ self.add_2216_bias = 0.0
+ self.conv2d_2218 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2220 = nn.ReLU()
+ self.module22_89 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_2225 = P.Split(axis=1, output_num=2)
+ self.add_2226_bias = 0.0
+ self.module20_88 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(13, 13),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_2237 = P.Reshape()
+ self.reshape_2237_shape = tuple([1, 1024, 1, 1])
+ self.split_2238 = P.Split(axis=1, output_num=2)
+ self.add_2241_bias = 0.0
+ self.conv2d_2243 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2245 = nn.ReLU()
+ self.avgpool2d_2246 = nn.AvgPool2d(kernel_size=(13, 13))
+ self.reshape_2247 = P.Reshape()
+ self.reshape_2247_shape = tuple([1, 2048])
+ self.flatten_2248 = nn.Flatten()
+ self.dense_2249 = nn.Dense(in_channels=2048, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ module22_0_opt = self.module22_0(opt_relu_1)
+ opt_maxpool2d_6 = self.pad_maxpool2d_6(module22_0_opt)
+ opt_maxpool2d_6 = self.maxpool2d_6(opt_maxpool2d_6)
+ module22_1_opt = self.module22_1(opt_maxpool2d_6)
+ opt_split_13, opt_split_13_1 = self.split_13(module22_1_opt)
+ opt_add_14 = opt_split_13 + self.add_14_bias
+ opt_add_15 = P.Add()(opt_add_14, opt_split_13_1)
+ module20_0_opt = self.module20_0(opt_add_15)
+ opt_reshape_25 = self.reshape_25(module20_0_opt, self.reshape_25_shape)
+ opt_split_26, opt_split_26_1 = self.split_26(opt_reshape_25)
+ opt_mul_27 = P.Mul()(opt_split_26, opt_split_13)
+ opt_mul_28 = P.Mul()(opt_split_26_1, opt_split_13_1)
+ opt_add_29 = opt_mul_27 + self.add_29_bias
+ opt_add_30 = P.Add()(opt_add_29, opt_mul_28)
+ opt_conv2d_31 = self.conv2d_31(opt_add_30)
+ opt_avgpool2d_8 = self.pad_avgpool2d_8(opt_maxpool2d_6)
+ opt_avgpool2d_8 = self.avgpool2d_8(opt_avgpool2d_8)
+ opt_conv2d_10 = self.conv2d_10(opt_avgpool2d_8)
+ opt_add_32 = P.Add()(opt_conv2d_31, opt_conv2d_10)
+ opt_relu_33 = self.relu_33(opt_add_32)
+ module22_2_opt = self.module22_2(opt_relu_33)
+ opt_split_38, opt_split_38_1 = self.split_38(module22_2_opt)
+ opt_add_39 = opt_split_38 + self.add_39_bias
+ opt_add_40 = P.Add()(opt_add_39, opt_split_38_1)
+ module20_1_opt = self.module20_1(opt_add_40)
+ opt_reshape_50 = self.reshape_50(module20_1_opt, self.reshape_50_shape)
+ opt_split_51, opt_split_51_1 = self.split_51(opt_reshape_50)
+ opt_mul_52 = P.Mul()(opt_split_51, opt_split_38)
+ opt_mul_53 = P.Mul()(opt_split_51_1, opt_split_38_1)
+ opt_add_54 = opt_mul_52 + self.add_54_bias
+ opt_add_55 = P.Add()(opt_add_54, opt_mul_53)
+ opt_conv2d_56 = self.conv2d_56(opt_add_55)
+ opt_add_57 = P.Add()(opt_conv2d_56, opt_relu_33)
+ opt_relu_58 = self.relu_58(opt_add_57)
+ module22_3_opt = self.module22_3(opt_relu_58)
+ opt_split_63, opt_split_63_1 = self.split_63(module22_3_opt)
+ opt_add_64 = opt_split_63 + self.add_64_bias
+ opt_add_65 = P.Add()(opt_add_64, opt_split_63_1)
+ module20_2_opt = self.module20_2(opt_add_65)
+ opt_reshape_75 = self.reshape_75(module20_2_opt, self.reshape_75_shape)
+ opt_split_76, opt_split_76_1 = self.split_76(opt_reshape_75)
+ opt_mul_77 = P.Mul()(opt_split_76, opt_split_63)
+ opt_mul_78 = P.Mul()(opt_split_76_1, opt_split_63_1)
+ opt_add_79 = opt_mul_77 + self.add_79_bias
+ opt_add_80 = P.Add()(opt_add_79, opt_mul_78)
+ opt_conv2d_81 = self.conv2d_81(opt_add_80)
+ opt_add_82 = P.Add()(opt_conv2d_81, opt_relu_58)
+ opt_relu_83 = self.relu_83(opt_add_82)
+ module22_4_opt = self.module22_4(opt_relu_83)
+ opt_split_90, opt_split_90_1 = self.split_90(module22_4_opt)
+ opt_add_91 = opt_split_90 + self.add_91_bias
+ opt_add_92 = P.Add()(opt_add_91, opt_split_90_1)
+ module20_3_opt = self.module20_3(opt_add_92)
+ opt_reshape_102 = self.reshape_102(module20_3_opt, self.reshape_102_shape)
+ opt_split_103, opt_split_103_1 = self.split_103(opt_reshape_102)
+ opt_mul_104 = P.Mul()(opt_split_103, opt_split_90)
+ opt_mul_105 = P.Mul()(opt_split_103_1, opt_split_90_1)
+ opt_add_106 = opt_mul_104 + self.add_106_bias
+ opt_add_107 = P.Add()(opt_add_106, opt_mul_105)
+ opt_pad_108 = self.pad_108(opt_add_107)
+ opt_avgpool2d_109 = self.pad_avgpool2d_109(opt_pad_108)
+ opt_avgpool2d_109 = self.avgpool2d_109(opt_avgpool2d_109)
+ opt_conv2d_110 = self.conv2d_110(opt_avgpool2d_109)
+ opt_avgpool2d_85 = self.pad_avgpool2d_85(opt_relu_83)
+ opt_avgpool2d_85 = self.avgpool2d_85(opt_avgpool2d_85)
+ opt_conv2d_87 = self.conv2d_87(opt_avgpool2d_85)
+ opt_add_111 = P.Add()(opt_conv2d_110, opt_conv2d_87)
+ opt_relu_112 = self.relu_112(opt_add_111)
+ module22_5_opt = self.module22_5(opt_relu_112)
+ opt_split_117, opt_split_117_1 = self.split_117(module22_5_opt)
+ opt_add_118 = opt_split_117 + self.add_118_bias
+ opt_add_119 = P.Add()(opt_add_118, opt_split_117_1)
+ module20_4_opt = self.module20_4(opt_add_119)
+ opt_reshape_129 = self.reshape_129(module20_4_opt, self.reshape_129_shape)
+ opt_split_130, opt_split_130_1 = self.split_130(opt_reshape_129)
+ opt_mul_131 = P.Mul()(opt_split_130, opt_split_117)
+ opt_mul_132 = P.Mul()(opt_split_130_1, opt_split_117_1)
+ opt_add_133 = opt_mul_131 + self.add_133_bias
+ opt_add_134 = P.Add()(opt_add_133, opt_mul_132)
+ opt_conv2d_135 = self.conv2d_135(opt_add_134)
+ opt_add_136 = P.Add()(opt_conv2d_135, opt_relu_112)
+ opt_relu_137 = self.relu_137(opt_add_136)
+ module22_6_opt = self.module22_6(opt_relu_137)
+ opt_split_142, opt_split_142_1 = self.split_142(module22_6_opt)
+ opt_add_143 = opt_split_142 + self.add_143_bias
+ opt_add_144 = P.Add()(opt_add_143, opt_split_142_1)
+ module20_5_opt = self.module20_5(opt_add_144)
+ opt_reshape_154 = self.reshape_154(module20_5_opt, self.reshape_154_shape)
+ opt_split_155, opt_split_155_1 = self.split_155(opt_reshape_154)
+ opt_mul_156 = P.Mul()(opt_split_155, opt_split_142)
+ opt_mul_157 = P.Mul()(opt_split_155_1, opt_split_142_1)
+ opt_add_158 = opt_mul_156 + self.add_158_bias
+ opt_add_159 = P.Add()(opt_add_158, opt_mul_157)
+ opt_conv2d_160 = self.conv2d_160(opt_add_159)
+ opt_add_161 = P.Add()(opt_conv2d_160, opt_relu_137)
+ opt_relu_162 = self.relu_162(opt_add_161)
+ module22_7_opt = self.module22_7(opt_relu_162)
+ opt_split_167, opt_split_167_1 = self.split_167(module22_7_opt)
+ opt_add_168 = opt_split_167 + self.add_168_bias
+ opt_add_169 = P.Add()(opt_add_168, opt_split_167_1)
+ module20_6_opt = self.module20_6(opt_add_169)
+ opt_reshape_179 = self.reshape_179(module20_6_opt, self.reshape_179_shape)
+ opt_split_180, opt_split_180_1 = self.split_180(opt_reshape_179)
+ opt_mul_181 = P.Mul()(opt_split_180, opt_split_167)
+ opt_mul_182 = P.Mul()(opt_split_180_1, opt_split_167_1)
+ opt_add_183 = opt_mul_181 + self.add_183_bias
+ opt_add_184 = P.Add()(opt_add_183, opt_mul_182)
+ opt_conv2d_185 = self.conv2d_185(opt_add_184)
+ opt_add_186 = P.Add()(opt_conv2d_185, opt_relu_162)
+ opt_relu_187 = self.relu_187(opt_add_186)
+ module22_8_opt = self.module22_8(opt_relu_187)
+ opt_split_192, opt_split_192_1 = self.split_192(module22_8_opt)
+ opt_add_193 = opt_split_192 + self.add_193_bias
+ opt_add_194 = P.Add()(opt_add_193, opt_split_192_1)
+ module20_7_opt = self.module20_7(opt_add_194)
+ opt_reshape_204 = self.reshape_204(module20_7_opt, self.reshape_204_shape)
+ opt_split_205, opt_split_205_1 = self.split_205(opt_reshape_204)
+ opt_mul_206 = P.Mul()(opt_split_205, opt_split_192)
+ opt_mul_207 = P.Mul()(opt_split_205_1, opt_split_192_1)
+ opt_add_208 = opt_mul_206 + self.add_208_bias
+ opt_add_209 = P.Add()(opt_add_208, opt_mul_207)
+ opt_conv2d_210 = self.conv2d_210(opt_add_209)
+ opt_add_211 = P.Add()(opt_conv2d_210, opt_relu_187)
+ opt_relu_212 = self.relu_212(opt_add_211)
+ module22_9_opt = self.module22_9(opt_relu_212)
+ opt_split_217, opt_split_217_1 = self.split_217(module22_9_opt)
+ opt_add_218 = opt_split_217 + self.add_218_bias
+ opt_add_219 = P.Add()(opt_add_218, opt_split_217_1)
+ module20_8_opt = self.module20_8(opt_add_219)
+ opt_reshape_229 = self.reshape_229(module20_8_opt, self.reshape_229_shape)
+ opt_split_230, opt_split_230_1 = self.split_230(opt_reshape_229)
+ opt_mul_231 = P.Mul()(opt_split_230, opt_split_217)
+ opt_mul_232 = P.Mul()(opt_split_230_1, opt_split_217_1)
+ opt_add_233 = opt_mul_231 + self.add_233_bias
+ opt_add_234 = P.Add()(opt_add_233, opt_mul_232)
+ opt_conv2d_235 = self.conv2d_235(opt_add_234)
+ opt_add_236 = P.Add()(opt_conv2d_235, opt_relu_212)
+ opt_relu_237 = self.relu_237(opt_add_236)
+ module22_10_opt = self.module22_10(opt_relu_237)
+ opt_split_242, opt_split_242_1 = self.split_242(module22_10_opt)
+ opt_add_243 = opt_split_242 + self.add_243_bias
+ opt_add_244 = P.Add()(opt_add_243, opt_split_242_1)
+ module20_9_opt = self.module20_9(opt_add_244)
+ opt_reshape_254 = self.reshape_254(module20_9_opt, self.reshape_254_shape)
+ opt_split_255, opt_split_255_1 = self.split_255(opt_reshape_254)
+ opt_mul_256 = P.Mul()(opt_split_255, opt_split_242)
+ opt_mul_257 = P.Mul()(opt_split_255_1, opt_split_242_1)
+ opt_add_258 = opt_mul_256 + self.add_258_bias
+ opt_add_259 = P.Add()(opt_add_258, opt_mul_257)
+ opt_conv2d_260 = self.conv2d_260(opt_add_259)
+ opt_add_261 = P.Add()(opt_conv2d_260, opt_relu_237)
+ opt_relu_262 = self.relu_262(opt_add_261)
+ module22_11_opt = self.module22_11(opt_relu_262)
+ opt_split_267, opt_split_267_1 = self.split_267(module22_11_opt)
+ opt_add_268 = opt_split_267 + self.add_268_bias
+ opt_add_269 = P.Add()(opt_add_268, opt_split_267_1)
+ module20_10_opt = self.module20_10(opt_add_269)
+ opt_reshape_279 = self.reshape_279(module20_10_opt, self.reshape_279_shape)
+ opt_split_280, opt_split_280_1 = self.split_280(opt_reshape_279)
+ opt_mul_281 = P.Mul()(opt_split_280, opt_split_267)
+ opt_mul_282 = P.Mul()(opt_split_280_1, opt_split_267_1)
+ opt_add_283 = opt_mul_281 + self.add_283_bias
+ opt_add_284 = P.Add()(opt_add_283, opt_mul_282)
+ opt_conv2d_285 = self.conv2d_285(opt_add_284)
+ opt_add_286 = P.Add()(opt_conv2d_285, opt_relu_262)
+ opt_relu_287 = self.relu_287(opt_add_286)
+ module22_12_opt = self.module22_12(opt_relu_287)
+ opt_split_292, opt_split_292_1 = self.split_292(module22_12_opt)
+ opt_add_293 = opt_split_292 + self.add_293_bias
+ opt_add_294 = P.Add()(opt_add_293, opt_split_292_1)
+ module20_11_opt = self.module20_11(opt_add_294)
+ opt_reshape_304 = self.reshape_304(module20_11_opt, self.reshape_304_shape)
+ opt_split_305, opt_split_305_1 = self.split_305(opt_reshape_304)
+ opt_mul_306 = P.Mul()(opt_split_305, opt_split_292)
+ opt_mul_307 = P.Mul()(opt_split_305_1, opt_split_292_1)
+ opt_add_308 = opt_mul_306 + self.add_308_bias
+ opt_add_309 = P.Add()(opt_add_308, opt_mul_307)
+ opt_conv2d_310 = self.conv2d_310(opt_add_309)
+ opt_add_311 = P.Add()(opt_conv2d_310, opt_relu_287)
+ opt_relu_312 = self.relu_312(opt_add_311)
+ module22_13_opt = self.module22_13(opt_relu_312)
+ opt_split_317, opt_split_317_1 = self.split_317(module22_13_opt)
+ opt_add_318 = opt_split_317 + self.add_318_bias
+ opt_add_319 = P.Add()(opt_add_318, opt_split_317_1)
+ module20_12_opt = self.module20_12(opt_add_319)
+ opt_reshape_329 = self.reshape_329(module20_12_opt, self.reshape_329_shape)
+ opt_split_330, opt_split_330_1 = self.split_330(opt_reshape_329)
+ opt_mul_331 = P.Mul()(opt_split_330, opt_split_317)
+ opt_mul_332 = P.Mul()(opt_split_330_1, opt_split_317_1)
+ opt_add_333 = opt_mul_331 + self.add_333_bias
+ opt_add_334 = P.Add()(opt_add_333, opt_mul_332)
+ opt_conv2d_335 = self.conv2d_335(opt_add_334)
+ opt_add_336 = P.Add()(opt_conv2d_335, opt_relu_312)
+ opt_relu_337 = self.relu_337(opt_add_336)
+ module22_14_opt = self.module22_14(opt_relu_337)
+ opt_split_342, opt_split_342_1 = self.split_342(module22_14_opt)
+ opt_add_343 = opt_split_342 + self.add_343_bias
+ opt_add_344 = P.Add()(opt_add_343, opt_split_342_1)
+ module20_13_opt = self.module20_13(opt_add_344)
+ opt_reshape_354 = self.reshape_354(module20_13_opt, self.reshape_354_shape)
+ opt_split_355, opt_split_355_1 = self.split_355(opt_reshape_354)
+ opt_mul_356 = P.Mul()(opt_split_355, opt_split_342)
+ opt_mul_357 = P.Mul()(opt_split_355_1, opt_split_342_1)
+ opt_add_358 = opt_mul_356 + self.add_358_bias
+ opt_add_359 = P.Add()(opt_add_358, opt_mul_357)
+ opt_conv2d_360 = self.conv2d_360(opt_add_359)
+ opt_add_361 = P.Add()(opt_conv2d_360, opt_relu_337)
+ opt_relu_362 = self.relu_362(opt_add_361)
+ module22_15_opt = self.module22_15(opt_relu_362)
+ opt_split_367, opt_split_367_1 = self.split_367(module22_15_opt)
+ opt_add_368 = opt_split_367 + self.add_368_bias
+ opt_add_369 = P.Add()(opt_add_368, opt_split_367_1)
+ module20_14_opt = self.module20_14(opt_add_369)
+ opt_reshape_379 = self.reshape_379(module20_14_opt, self.reshape_379_shape)
+ opt_split_380, opt_split_380_1 = self.split_380(opt_reshape_379)
+ opt_mul_381 = P.Mul()(opt_split_380, opt_split_367)
+ opt_mul_382 = P.Mul()(opt_split_380_1, opt_split_367_1)
+ opt_add_383 = opt_mul_381 + self.add_383_bias
+ opt_add_384 = P.Add()(opt_add_383, opt_mul_382)
+ opt_conv2d_385 = self.conv2d_385(opt_add_384)
+ opt_add_386 = P.Add()(opt_conv2d_385, opt_relu_362)
+ opt_relu_387 = self.relu_387(opt_add_386)
+ module22_16_opt = self.module22_16(opt_relu_387)
+ opt_split_392, opt_split_392_1 = self.split_392(module22_16_opt)
+ opt_add_393 = opt_split_392 + self.add_393_bias
+ opt_add_394 = P.Add()(opt_add_393, opt_split_392_1)
+ module20_15_opt = self.module20_15(opt_add_394)
+ opt_reshape_404 = self.reshape_404(module20_15_opt, self.reshape_404_shape)
+ opt_split_405, opt_split_405_1 = self.split_405(opt_reshape_404)
+ opt_mul_406 = P.Mul()(opt_split_405, opt_split_392)
+ opt_mul_407 = P.Mul()(opt_split_405_1, opt_split_392_1)
+ opt_add_408 = opt_mul_406 + self.add_408_bias
+ opt_add_409 = P.Add()(opt_add_408, opt_mul_407)
+ opt_conv2d_410 = self.conv2d_410(opt_add_409)
+ opt_add_411 = P.Add()(opt_conv2d_410, opt_relu_387)
+ opt_relu_412 = self.relu_412(opt_add_411)
+ module22_17_opt = self.module22_17(opt_relu_412)
+ opt_split_417, opt_split_417_1 = self.split_417(module22_17_opt)
+ opt_add_418 = opt_split_417 + self.add_418_bias
+ opt_add_419 = P.Add()(opt_add_418, opt_split_417_1)
+ module20_16_opt = self.module20_16(opt_add_419)
+ opt_reshape_429 = self.reshape_429(module20_16_opt, self.reshape_429_shape)
+ opt_split_430, opt_split_430_1 = self.split_430(opt_reshape_429)
+ opt_mul_431 = P.Mul()(opt_split_430, opt_split_417)
+ opt_mul_432 = P.Mul()(opt_split_430_1, opt_split_417_1)
+ opt_add_433 = opt_mul_431 + self.add_433_bias
+ opt_add_434 = P.Add()(opt_add_433, opt_mul_432)
+ opt_conv2d_435 = self.conv2d_435(opt_add_434)
+ opt_add_436 = P.Add()(opt_conv2d_435, opt_relu_412)
+ opt_relu_437 = self.relu_437(opt_add_436)
+ module22_18_opt = self.module22_18(opt_relu_437)
+ opt_split_442, opt_split_442_1 = self.split_442(module22_18_opt)
+ opt_add_443 = opt_split_442 + self.add_443_bias
+ opt_add_444 = P.Add()(opt_add_443, opt_split_442_1)
+ module20_17_opt = self.module20_17(opt_add_444)
+ opt_reshape_454 = self.reshape_454(module20_17_opt, self.reshape_454_shape)
+ opt_split_455, opt_split_455_1 = self.split_455(opt_reshape_454)
+ opt_mul_456 = P.Mul()(opt_split_455, opt_split_442)
+ opt_mul_457 = P.Mul()(opt_split_455_1, opt_split_442_1)
+ opt_add_458 = opt_mul_456 + self.add_458_bias
+ opt_add_459 = P.Add()(opt_add_458, opt_mul_457)
+ opt_conv2d_460 = self.conv2d_460(opt_add_459)
+ opt_add_461 = P.Add()(opt_conv2d_460, opt_relu_437)
+ opt_relu_462 = self.relu_462(opt_add_461)
+ module22_19_opt = self.module22_19(opt_relu_462)
+ opt_split_467, opt_split_467_1 = self.split_467(module22_19_opt)
+ opt_add_468 = opt_split_467 + self.add_468_bias
+ opt_add_469 = P.Add()(opt_add_468, opt_split_467_1)
+ module20_18_opt = self.module20_18(opt_add_469)
+ opt_reshape_479 = self.reshape_479(module20_18_opt, self.reshape_479_shape)
+ opt_split_480, opt_split_480_1 = self.split_480(opt_reshape_479)
+ opt_mul_481 = P.Mul()(opt_split_480, opt_split_467)
+ opt_mul_482 = P.Mul()(opt_split_480_1, opt_split_467_1)
+ opt_add_483 = opt_mul_481 + self.add_483_bias
+ opt_add_484 = P.Add()(opt_add_483, opt_mul_482)
+ opt_conv2d_485 = self.conv2d_485(opt_add_484)
+ opt_add_486 = P.Add()(opt_conv2d_485, opt_relu_462)
+ opt_relu_487 = self.relu_487(opt_add_486)
+ module22_20_opt = self.module22_20(opt_relu_487)
+ opt_split_492, opt_split_492_1 = self.split_492(module22_20_opt)
+ opt_add_493 = opt_split_492 + self.add_493_bias
+ opt_add_494 = P.Add()(opt_add_493, opt_split_492_1)
+ module20_19_opt = self.module20_19(opt_add_494)
+ opt_reshape_504 = self.reshape_504(module20_19_opt, self.reshape_504_shape)
+ opt_split_505, opt_split_505_1 = self.split_505(opt_reshape_504)
+ opt_mul_506 = P.Mul()(opt_split_505, opt_split_492)
+ opt_mul_507 = P.Mul()(opt_split_505_1, opt_split_492_1)
+ opt_add_508 = opt_mul_506 + self.add_508_bias
+ opt_add_509 = P.Add()(opt_add_508, opt_mul_507)
+ opt_conv2d_510 = self.conv2d_510(opt_add_509)
+ opt_add_511 = P.Add()(opt_conv2d_510, opt_relu_487)
+ opt_relu_512 = self.relu_512(opt_add_511)
+ module22_21_opt = self.module22_21(opt_relu_512)
+ opt_split_517, opt_split_517_1 = self.split_517(module22_21_opt)
+ opt_add_518 = opt_split_517 + self.add_518_bias
+ opt_add_519 = P.Add()(opt_add_518, opt_split_517_1)
+ module20_20_opt = self.module20_20(opt_add_519)
+ opt_reshape_529 = self.reshape_529(module20_20_opt, self.reshape_529_shape)
+ opt_split_530, opt_split_530_1 = self.split_530(opt_reshape_529)
+ opt_mul_531 = P.Mul()(opt_split_530, opt_split_517)
+ opt_mul_532 = P.Mul()(opt_split_530_1, opt_split_517_1)
+ opt_add_533 = opt_mul_531 + self.add_533_bias
+ opt_add_534 = P.Add()(opt_add_533, opt_mul_532)
+ opt_conv2d_535 = self.conv2d_535(opt_add_534)
+ opt_add_536 = P.Add()(opt_conv2d_535, opt_relu_512)
+ opt_relu_537 = self.relu_537(opt_add_536)
+ module22_22_opt = self.module22_22(opt_relu_537)
+ opt_split_542, opt_split_542_1 = self.split_542(module22_22_opt)
+ opt_add_543 = opt_split_542 + self.add_543_bias
+ opt_add_544 = P.Add()(opt_add_543, opt_split_542_1)
+ module20_21_opt = self.module20_21(opt_add_544)
+ opt_reshape_554 = self.reshape_554(module20_21_opt, self.reshape_554_shape)
+ opt_split_555, opt_split_555_1 = self.split_555(opt_reshape_554)
+ opt_mul_556 = P.Mul()(opt_split_555, opt_split_542)
+ opt_mul_557 = P.Mul()(opt_split_555_1, opt_split_542_1)
+ opt_add_558 = opt_mul_556 + self.add_558_bias
+ opt_add_559 = P.Add()(opt_add_558, opt_mul_557)
+ opt_conv2d_560 = self.conv2d_560(opt_add_559)
+ opt_add_561 = P.Add()(opt_conv2d_560, opt_relu_537)
+ opt_relu_562 = self.relu_562(opt_add_561)
+ module22_23_opt = self.module22_23(opt_relu_562)
+ opt_split_567, opt_split_567_1 = self.split_567(module22_23_opt)
+ opt_add_568 = opt_split_567 + self.add_568_bias
+ opt_add_569 = P.Add()(opt_add_568, opt_split_567_1)
+ module20_22_opt = self.module20_22(opt_add_569)
+ opt_reshape_579 = self.reshape_579(module20_22_opt, self.reshape_579_shape)
+ opt_split_580, opt_split_580_1 = self.split_580(opt_reshape_579)
+ opt_mul_581 = P.Mul()(opt_split_580, opt_split_567)
+ opt_mul_582 = P.Mul()(opt_split_580_1, opt_split_567_1)
+ opt_add_583 = opt_mul_581 + self.add_583_bias
+ opt_add_584 = P.Add()(opt_add_583, opt_mul_582)
+ opt_conv2d_585 = self.conv2d_585(opt_add_584)
+ opt_add_586 = P.Add()(opt_conv2d_585, opt_relu_562)
+ opt_relu_587 = self.relu_587(opt_add_586)
+ module22_24_opt = self.module22_24(opt_relu_587)
+ opt_split_592, opt_split_592_1 = self.split_592(module22_24_opt)
+ opt_add_593 = opt_split_592 + self.add_593_bias
+ opt_add_594 = P.Add()(opt_add_593, opt_split_592_1)
+ module20_23_opt = self.module20_23(opt_add_594)
+ opt_reshape_604 = self.reshape_604(module20_23_opt, self.reshape_604_shape)
+ opt_split_605, opt_split_605_1 = self.split_605(opt_reshape_604)
+ opt_mul_606 = P.Mul()(opt_split_605, opt_split_592)
+ opt_mul_607 = P.Mul()(opt_split_605_1, opt_split_592_1)
+ opt_add_608 = opt_mul_606 + self.add_608_bias
+ opt_add_609 = P.Add()(opt_add_608, opt_mul_607)
+ opt_conv2d_610 = self.conv2d_610(opt_add_609)
+ opt_add_611 = P.Add()(opt_conv2d_610, opt_relu_587)
+ opt_relu_612 = self.relu_612(opt_add_611)
+ module22_25_opt = self.module22_25(opt_relu_612)
+ opt_split_617, opt_split_617_1 = self.split_617(module22_25_opt)
+ opt_add_618 = opt_split_617 + self.add_618_bias
+ opt_add_619 = P.Add()(opt_add_618, opt_split_617_1)
+ module20_24_opt = self.module20_24(opt_add_619)
+ opt_reshape_629 = self.reshape_629(module20_24_opt, self.reshape_629_shape)
+ opt_split_630, opt_split_630_1 = self.split_630(opt_reshape_629)
+ opt_mul_631 = P.Mul()(opt_split_630, opt_split_617)
+ opt_mul_632 = P.Mul()(opt_split_630_1, opt_split_617_1)
+ opt_add_633 = opt_mul_631 + self.add_633_bias
+ opt_add_634 = P.Add()(opt_add_633, opt_mul_632)
+ opt_conv2d_635 = self.conv2d_635(opt_add_634)
+ opt_add_636 = P.Add()(opt_conv2d_635, opt_relu_612)
+ opt_relu_637 = self.relu_637(opt_add_636)
+ module22_26_opt = self.module22_26(opt_relu_637)
+ opt_split_642, opt_split_642_1 = self.split_642(module22_26_opt)
+ opt_add_643 = opt_split_642 + self.add_643_bias
+ opt_add_644 = P.Add()(opt_add_643, opt_split_642_1)
+ module20_25_opt = self.module20_25(opt_add_644)
+ opt_reshape_654 = self.reshape_654(module20_25_opt, self.reshape_654_shape)
+ opt_split_655, opt_split_655_1 = self.split_655(opt_reshape_654)
+ opt_mul_656 = P.Mul()(opt_split_655, opt_split_642)
+ opt_mul_657 = P.Mul()(opt_split_655_1, opt_split_642_1)
+ opt_add_658 = opt_mul_656 + self.add_658_bias
+ opt_add_659 = P.Add()(opt_add_658, opt_mul_657)
+ opt_conv2d_660 = self.conv2d_660(opt_add_659)
+ opt_add_661 = P.Add()(opt_conv2d_660, opt_relu_637)
+ opt_relu_662 = self.relu_662(opt_add_661)
+ module22_27_opt = self.module22_27(opt_relu_662)
+ opt_split_667, opt_split_667_1 = self.split_667(module22_27_opt)
+ opt_add_668 = opt_split_667 + self.add_668_bias
+ opt_add_669 = P.Add()(opt_add_668, opt_split_667_1)
+ module20_26_opt = self.module20_26(opt_add_669)
+ opt_reshape_679 = self.reshape_679(module20_26_opt, self.reshape_679_shape)
+ opt_split_680, opt_split_680_1 = self.split_680(opt_reshape_679)
+ opt_mul_681 = P.Mul()(opt_split_680, opt_split_667)
+ opt_mul_682 = P.Mul()(opt_split_680_1, opt_split_667_1)
+ opt_add_683 = opt_mul_681 + self.add_683_bias
+ opt_add_684 = P.Add()(opt_add_683, opt_mul_682)
+ opt_conv2d_685 = self.conv2d_685(opt_add_684)
+ opt_add_686 = P.Add()(opt_conv2d_685, opt_relu_662)
+ opt_relu_687 = self.relu_687(opt_add_686)
+ module22_28_opt = self.module22_28(opt_relu_687)
+ opt_split_692, opt_split_692_1 = self.split_692(module22_28_opt)
+ opt_add_693 = opt_split_692 + self.add_693_bias
+ opt_add_694 = P.Add()(opt_add_693, opt_split_692_1)
+ module20_27_opt = self.module20_27(opt_add_694)
+ opt_reshape_704 = self.reshape_704(module20_27_opt, self.reshape_704_shape)
+ opt_split_705, opt_split_705_1 = self.split_705(opt_reshape_704)
+ opt_mul_706 = P.Mul()(opt_split_705, opt_split_692)
+ opt_mul_707 = P.Mul()(opt_split_705_1, opt_split_692_1)
+ opt_add_708 = opt_mul_706 + self.add_708_bias
+ opt_add_709 = P.Add()(opt_add_708, opt_mul_707)
+ opt_conv2d_710 = self.conv2d_710(opt_add_709)
+ opt_add_711 = P.Add()(opt_conv2d_710, opt_relu_687)
+ opt_relu_712 = self.relu_712(opt_add_711)
+ module22_29_opt = self.module22_29(opt_relu_712)
+ opt_split_717, opt_split_717_1 = self.split_717(module22_29_opt)
+ opt_add_718 = opt_split_717 + self.add_718_bias
+ opt_add_719 = P.Add()(opt_add_718, opt_split_717_1)
+ module20_28_opt = self.module20_28(opt_add_719)
+ opt_reshape_729 = self.reshape_729(module20_28_opt, self.reshape_729_shape)
+ opt_split_730, opt_split_730_1 = self.split_730(opt_reshape_729)
+ opt_mul_731 = P.Mul()(opt_split_730, opt_split_717)
+ opt_mul_732 = P.Mul()(opt_split_730_1, opt_split_717_1)
+ opt_add_733 = opt_mul_731 + self.add_733_bias
+ opt_add_734 = P.Add()(opt_add_733, opt_mul_732)
+ opt_conv2d_735 = self.conv2d_735(opt_add_734)
+ opt_add_736 = P.Add()(opt_conv2d_735, opt_relu_712)
+ opt_relu_737 = self.relu_737(opt_add_736)
+ module22_30_opt = self.module22_30(opt_relu_737)
+ opt_split_742, opt_split_742_1 = self.split_742(module22_30_opt)
+ opt_add_743 = opt_split_742 + self.add_743_bias
+ opt_add_744 = P.Add()(opt_add_743, opt_split_742_1)
+ module20_29_opt = self.module20_29(opt_add_744)
+ opt_reshape_754 = self.reshape_754(module20_29_opt, self.reshape_754_shape)
+ opt_split_755, opt_split_755_1 = self.split_755(opt_reshape_754)
+ opt_mul_756 = P.Mul()(opt_split_755, opt_split_742)
+ opt_mul_757 = P.Mul()(opt_split_755_1, opt_split_742_1)
+ opt_add_758 = opt_mul_756 + self.add_758_bias
+ opt_add_759 = P.Add()(opt_add_758, opt_mul_757)
+ opt_conv2d_760 = self.conv2d_760(opt_add_759)
+ opt_add_761 = P.Add()(opt_conv2d_760, opt_relu_737)
+ opt_relu_762 = self.relu_762(opt_add_761)
+ module22_31_opt = self.module22_31(opt_relu_762)
+ opt_split_767, opt_split_767_1 = self.split_767(module22_31_opt)
+ opt_add_768 = opt_split_767 + self.add_768_bias
+ opt_add_769 = P.Add()(opt_add_768, opt_split_767_1)
+ module20_30_opt = self.module20_30(opt_add_769)
+ opt_reshape_779 = self.reshape_779(module20_30_opt, self.reshape_779_shape)
+ opt_split_780, opt_split_780_1 = self.split_780(opt_reshape_779)
+ opt_mul_781 = P.Mul()(opt_split_780, opt_split_767)
+ opt_mul_782 = P.Mul()(opt_split_780_1, opt_split_767_1)
+ opt_add_783 = opt_mul_781 + self.add_783_bias
+ opt_add_784 = P.Add()(opt_add_783, opt_mul_782)
+ opt_conv2d_785 = self.conv2d_785(opt_add_784)
+ opt_add_786 = P.Add()(opt_conv2d_785, opt_relu_762)
+ opt_relu_787 = self.relu_787(opt_add_786)
+ module22_32_opt = self.module22_32(opt_relu_787)
+ opt_split_792, opt_split_792_1 = self.split_792(module22_32_opt)
+ opt_add_793 = opt_split_792 + self.add_793_bias
+ opt_add_794 = P.Add()(opt_add_793, opt_split_792_1)
+ module20_31_opt = self.module20_31(opt_add_794)
+ opt_reshape_804 = self.reshape_804(module20_31_opt, self.reshape_804_shape)
+ opt_split_805, opt_split_805_1 = self.split_805(opt_reshape_804)
+ opt_mul_806 = P.Mul()(opt_split_805, opt_split_792)
+ opt_mul_807 = P.Mul()(opt_split_805_1, opt_split_792_1)
+ opt_add_808 = opt_mul_806 + self.add_808_bias
+ opt_add_809 = P.Add()(opt_add_808, opt_mul_807)
+ opt_conv2d_810 = self.conv2d_810(opt_add_809)
+ opt_add_811 = P.Add()(opt_conv2d_810, opt_relu_787)
+ opt_relu_812 = self.relu_812(opt_add_811)
+ module22_33_opt = self.module22_33(opt_relu_812)
+ opt_split_817, opt_split_817_1 = self.split_817(module22_33_opt)
+ opt_add_818 = opt_split_817 + self.add_818_bias
+ opt_add_819 = P.Add()(opt_add_818, opt_split_817_1)
+ module20_32_opt = self.module20_32(opt_add_819)
+ opt_reshape_829 = self.reshape_829(module20_32_opt, self.reshape_829_shape)
+ opt_split_830, opt_split_830_1 = self.split_830(opt_reshape_829)
+ opt_mul_831 = P.Mul()(opt_split_830, opt_split_817)
+ opt_mul_832 = P.Mul()(opt_split_830_1, opt_split_817_1)
+ opt_add_833 = opt_mul_831 + self.add_833_bias
+ opt_add_834 = P.Add()(opt_add_833, opt_mul_832)
+ opt_conv2d_835 = self.conv2d_835(opt_add_834)
+ opt_add_836 = P.Add()(opt_conv2d_835, opt_relu_812)
+ opt_relu_837 = self.relu_837(opt_add_836)
+ module22_34_opt = self.module22_34(opt_relu_837)
+ opt_split_844, opt_split_844_1 = self.split_844(module22_34_opt)
+ opt_add_845 = opt_split_844 + self.add_845_bias
+ opt_add_846 = P.Add()(opt_add_845, opt_split_844_1)
+ module20_33_opt = self.module20_33(opt_add_846)
+ opt_reshape_856 = self.reshape_856(module20_33_opt, self.reshape_856_shape)
+ opt_split_857, opt_split_857_1 = self.split_857(opt_reshape_856)
+ opt_mul_858 = P.Mul()(opt_split_857, opt_split_844)
+ opt_mul_859 = P.Mul()(opt_split_857_1, opt_split_844_1)
+ opt_add_860 = opt_mul_858 + self.add_860_bias
+ opt_add_861 = P.Add()(opt_add_860, opt_mul_859)
+ opt_pad_862 = self.pad_862(opt_add_861)
+ opt_avgpool2d_863 = self.pad_avgpool2d_863(opt_pad_862)
+ opt_avgpool2d_863 = self.avgpool2d_863(opt_avgpool2d_863)
+ opt_conv2d_864 = self.conv2d_864(opt_avgpool2d_863)
+ opt_avgpool2d_839 = self.pad_avgpool2d_839(opt_relu_837)
+ opt_avgpool2d_839 = self.avgpool2d_839(opt_avgpool2d_839)
+ opt_conv2d_841 = self.conv2d_841(opt_avgpool2d_839)
+ opt_add_865 = P.Add()(opt_conv2d_864, opt_conv2d_841)
+ opt_relu_866 = self.relu_866(opt_add_865)
+ module22_35_opt = self.module22_35(opt_relu_866)
+ opt_split_871, opt_split_871_1 = self.split_871(module22_35_opt)
+ opt_add_872 = opt_split_871 + self.add_872_bias
+ opt_add_873 = P.Add()(opt_add_872, opt_split_871_1)
+ module20_34_opt = self.module20_34(opt_add_873)
+ opt_reshape_883 = self.reshape_883(module20_34_opt, self.reshape_883_shape)
+ opt_split_884, opt_split_884_1 = self.split_884(opt_reshape_883)
+ opt_mul_885 = P.Mul()(opt_split_884, opt_split_871)
+ opt_mul_886 = P.Mul()(opt_split_884_1, opt_split_871_1)
+ opt_add_887 = opt_mul_885 + self.add_887_bias
+ opt_add_888 = P.Add()(opt_add_887, opt_mul_886)
+ opt_conv2d_889 = self.conv2d_889(opt_add_888)
+ opt_add_890 = P.Add()(opt_conv2d_889, opt_relu_866)
+ opt_relu_891 = self.relu_891(opt_add_890)
+ module22_36_opt = self.module22_36(opt_relu_891)
+ opt_split_896, opt_split_896_1 = self.split_896(module22_36_opt)
+ opt_add_897 = opt_split_896 + self.add_897_bias
+ opt_add_898 = P.Add()(opt_add_897, opt_split_896_1)
+ module20_35_opt = self.module20_35(opt_add_898)
+ opt_reshape_908 = self.reshape_908(module20_35_opt, self.reshape_908_shape)
+ opt_split_909, opt_split_909_1 = self.split_909(opt_reshape_908)
+ opt_mul_910 = P.Mul()(opt_split_909, opt_split_896)
+ opt_mul_911 = P.Mul()(opt_split_909_1, opt_split_896_1)
+ opt_add_912 = opt_mul_910 + self.add_912_bias
+ opt_add_913 = P.Add()(opt_add_912, opt_mul_911)
+ opt_conv2d_914 = self.conv2d_914(opt_add_913)
+ opt_add_915 = P.Add()(opt_conv2d_914, opt_relu_891)
+ opt_relu_916 = self.relu_916(opt_add_915)
+ module22_37_opt = self.module22_37(opt_relu_916)
+ opt_split_921, opt_split_921_1 = self.split_921(module22_37_opt)
+ opt_add_922 = opt_split_921 + self.add_922_bias
+ opt_add_923 = P.Add()(opt_add_922, opt_split_921_1)
+ module20_36_opt = self.module20_36(opt_add_923)
+ opt_reshape_933 = self.reshape_933(module20_36_opt, self.reshape_933_shape)
+ opt_split_934, opt_split_934_1 = self.split_934(opt_reshape_933)
+ opt_mul_935 = P.Mul()(opt_split_934, opt_split_921)
+ opt_mul_936 = P.Mul()(opt_split_934_1, opt_split_921_1)
+ opt_add_937 = opt_mul_935 + self.add_937_bias
+ opt_add_938 = P.Add()(opt_add_937, opt_mul_936)
+ opt_conv2d_939 = self.conv2d_939(opt_add_938)
+ opt_add_940 = P.Add()(opt_conv2d_939, opt_relu_916)
+ opt_relu_941 = self.relu_941(opt_add_940)
+ module22_38_opt = self.module22_38(opt_relu_941)
+ opt_split_946, opt_split_946_1 = self.split_946(module22_38_opt)
+ opt_add_947 = opt_split_946 + self.add_947_bias
+ opt_add_948 = P.Add()(opt_add_947, opt_split_946_1)
+ module20_37_opt = self.module20_37(opt_add_948)
+ opt_reshape_958 = self.reshape_958(module20_37_opt, self.reshape_958_shape)
+ opt_split_959, opt_split_959_1 = self.split_959(opt_reshape_958)
+ opt_mul_960 = P.Mul()(opt_split_959, opt_split_946)
+ opt_mul_961 = P.Mul()(opt_split_959_1, opt_split_946_1)
+ opt_add_962 = opt_mul_960 + self.add_962_bias
+ opt_add_963 = P.Add()(opt_add_962, opt_mul_961)
+ opt_conv2d_964 = self.conv2d_964(opt_add_963)
+ opt_add_965 = P.Add()(opt_conv2d_964, opt_relu_941)
+ opt_relu_966 = self.relu_966(opt_add_965)
+ module22_39_opt = self.module22_39(opt_relu_966)
+ opt_split_971, opt_split_971_1 = self.split_971(module22_39_opt)
+ opt_add_972 = opt_split_971 + self.add_972_bias
+ opt_add_973 = P.Add()(opt_add_972, opt_split_971_1)
+ module20_38_opt = self.module20_38(opt_add_973)
+ opt_reshape_983 = self.reshape_983(module20_38_opt, self.reshape_983_shape)
+ opt_split_984, opt_split_984_1 = self.split_984(opt_reshape_983)
+ opt_mul_985 = P.Mul()(opt_split_984, opt_split_971)
+ opt_mul_986 = P.Mul()(opt_split_984_1, opt_split_971_1)
+ opt_add_987 = opt_mul_985 + self.add_987_bias
+ opt_add_988 = P.Add()(opt_add_987, opt_mul_986)
+ opt_conv2d_989 = self.conv2d_989(opt_add_988)
+ opt_add_990 = P.Add()(opt_conv2d_989, opt_relu_966)
+ opt_relu_991 = self.relu_991(opt_add_990)
+ module22_40_opt = self.module22_40(opt_relu_991)
+ opt_split_996, opt_split_996_1 = self.split_996(module22_40_opt)
+ opt_add_997 = opt_split_996 + self.add_997_bias
+ opt_add_998 = P.Add()(opt_add_997, opt_split_996_1)
+ module20_39_opt = self.module20_39(opt_add_998)
+ opt_reshape_1008 = self.reshape_1008(module20_39_opt, self.reshape_1008_shape)
+ opt_split_1009, opt_split_1009_1 = self.split_1009(opt_reshape_1008)
+ opt_mul_1010 = P.Mul()(opt_split_1009, opt_split_996)
+ opt_mul_1011 = P.Mul()(opt_split_1009_1, opt_split_996_1)
+ opt_add_1012 = opt_mul_1010 + self.add_1012_bias
+ opt_add_1013 = P.Add()(opt_add_1012, opt_mul_1011)
+ opt_conv2d_1014 = self.conv2d_1014(opt_add_1013)
+ opt_add_1015 = P.Add()(opt_conv2d_1014, opt_relu_991)
+ opt_relu_1016 = self.relu_1016(opt_add_1015)
+ module22_41_opt = self.module22_41(opt_relu_1016)
+ opt_split_1021, opt_split_1021_1 = self.split_1021(module22_41_opt)
+ opt_add_1022 = opt_split_1021 + self.add_1022_bias
+ opt_add_1023 = P.Add()(opt_add_1022, opt_split_1021_1)
+ module20_40_opt = self.module20_40(opt_add_1023)
+ opt_reshape_1033 = self.reshape_1033(module20_40_opt, self.reshape_1033_shape)
+ opt_split_1034, opt_split_1034_1 = self.split_1034(opt_reshape_1033)
+ opt_mul_1035 = P.Mul()(opt_split_1034, opt_split_1021)
+ opt_mul_1036 = P.Mul()(opt_split_1034_1, opt_split_1021_1)
+ opt_add_1037 = opt_mul_1035 + self.add_1037_bias
+ opt_add_1038 = P.Add()(opt_add_1037, opt_mul_1036)
+ opt_conv2d_1039 = self.conv2d_1039(opt_add_1038)
+ opt_add_1040 = P.Add()(opt_conv2d_1039, opt_relu_1016)
+ opt_relu_1041 = self.relu_1041(opt_add_1040)
+ module22_42_opt = self.module22_42(opt_relu_1041)
+ opt_split_1046, opt_split_1046_1 = self.split_1046(module22_42_opt)
+ opt_add_1047 = opt_split_1046 + self.add_1047_bias
+ opt_add_1048 = P.Add()(opt_add_1047, opt_split_1046_1)
+ module20_41_opt = self.module20_41(opt_add_1048)
+ opt_reshape_1058 = self.reshape_1058(module20_41_opt, self.reshape_1058_shape)
+ opt_split_1059, opt_split_1059_1 = self.split_1059(opt_reshape_1058)
+ opt_mul_1060 = P.Mul()(opt_split_1059, opt_split_1046)
+ opt_mul_1061 = P.Mul()(opt_split_1059_1, opt_split_1046_1)
+ opt_add_1062 = opt_mul_1060 + self.add_1062_bias
+ opt_add_1063 = P.Add()(opt_add_1062, opt_mul_1061)
+ opt_conv2d_1064 = self.conv2d_1064(opt_add_1063)
+ opt_add_1065 = P.Add()(opt_conv2d_1064, opt_relu_1041)
+ opt_relu_1066 = self.relu_1066(opt_add_1065)
+ module22_43_opt = self.module22_43(opt_relu_1066)
+ opt_split_1071, opt_split_1071_1 = self.split_1071(module22_43_opt)
+ opt_add_1072 = opt_split_1071 + self.add_1072_bias
+ opt_add_1073 = P.Add()(opt_add_1072, opt_split_1071_1)
+ module20_42_opt = self.module20_42(opt_add_1073)
+ opt_reshape_1083 = self.reshape_1083(module20_42_opt, self.reshape_1083_shape)
+ opt_split_1084, opt_split_1084_1 = self.split_1084(opt_reshape_1083)
+ opt_mul_1085 = P.Mul()(opt_split_1084, opt_split_1071)
+ opt_mul_1086 = P.Mul()(opt_split_1084_1, opt_split_1071_1)
+ opt_add_1087 = opt_mul_1085 + self.add_1087_bias
+ opt_add_1088 = P.Add()(opt_add_1087, opt_mul_1086)
+ opt_conv2d_1089 = self.conv2d_1089(opt_add_1088)
+ opt_add_1090 = P.Add()(opt_conv2d_1089, opt_relu_1066)
+ opt_relu_1091 = self.relu_1091(opt_add_1090)
+ module22_44_opt = self.module22_44(opt_relu_1091)
+ opt_split_1096, opt_split_1096_1 = self.split_1096(module22_44_opt)
+ opt_add_1097 = opt_split_1096 + self.add_1097_bias
+ opt_add_1098 = P.Add()(opt_add_1097, opt_split_1096_1)
+ module20_43_opt = self.module20_43(opt_add_1098)
+ opt_reshape_1108 = self.reshape_1108(module20_43_opt, self.reshape_1108_shape)
+ opt_split_1109, opt_split_1109_1 = self.split_1109(opt_reshape_1108)
+ opt_mul_1110 = P.Mul()(opt_split_1109, opt_split_1096)
+ opt_mul_1111 = P.Mul()(opt_split_1109_1, opt_split_1096_1)
+ opt_add_1112 = opt_mul_1110 + self.add_1112_bias
+ opt_add_1113 = P.Add()(opt_add_1112, opt_mul_1111)
+ opt_conv2d_1114 = self.conv2d_1114(opt_add_1113)
+ opt_add_1115 = P.Add()(opt_conv2d_1114, opt_relu_1091)
+ opt_relu_1116 = self.relu_1116(opt_add_1115)
+ module22_45_opt = self.module22_45(opt_relu_1116)
+ opt_split_1121, opt_split_1121_1 = self.split_1121(module22_45_opt)
+ opt_add_1122 = opt_split_1121 + self.add_1122_bias
+ opt_add_1123 = P.Add()(opt_add_1122, opt_split_1121_1)
+ module20_44_opt = self.module20_44(opt_add_1123)
+ opt_reshape_1133 = self.reshape_1133(module20_44_opt, self.reshape_1133_shape)
+ opt_split_1134, opt_split_1134_1 = self.split_1134(opt_reshape_1133)
+ opt_mul_1135 = P.Mul()(opt_split_1134, opt_split_1121)
+ opt_mul_1136 = P.Mul()(opt_split_1134_1, opt_split_1121_1)
+ opt_add_1137 = opt_mul_1135 + self.add_1137_bias
+ opt_add_1138 = P.Add()(opt_add_1137, opt_mul_1136)
+ opt_conv2d_1139 = self.conv2d_1139(opt_add_1138)
+ opt_add_1140 = P.Add()(opt_conv2d_1139, opt_relu_1116)
+ opt_relu_1141 = self.relu_1141(opt_add_1140)
+ module22_46_opt = self.module22_46(opt_relu_1141)
+ opt_split_1146, opt_split_1146_1 = self.split_1146(module22_46_opt)
+ opt_add_1147 = opt_split_1146 + self.add_1147_bias
+ opt_add_1148 = P.Add()(opt_add_1147, opt_split_1146_1)
+ module20_45_opt = self.module20_45(opt_add_1148)
+ opt_reshape_1158 = self.reshape_1158(module20_45_opt, self.reshape_1158_shape)
+ opt_split_1159, opt_split_1159_1 = self.split_1159(opt_reshape_1158)
+ opt_mul_1160 = P.Mul()(opt_split_1159, opt_split_1146)
+ opt_mul_1161 = P.Mul()(opt_split_1159_1, opt_split_1146_1)
+ opt_add_1162 = opt_mul_1160 + self.add_1162_bias
+ opt_add_1163 = P.Add()(opt_add_1162, opt_mul_1161)
+ opt_conv2d_1164 = self.conv2d_1164(opt_add_1163)
+ opt_add_1165 = P.Add()(opt_conv2d_1164, opt_relu_1141)
+ opt_relu_1166 = self.relu_1166(opt_add_1165)
+ module22_47_opt = self.module22_47(opt_relu_1166)
+ opt_split_1171, opt_split_1171_1 = self.split_1171(module22_47_opt)
+ opt_add_1172 = opt_split_1171 + self.add_1172_bias
+ opt_add_1173 = P.Add()(opt_add_1172, opt_split_1171_1)
+ module20_46_opt = self.module20_46(opt_add_1173)
+ opt_reshape_1183 = self.reshape_1183(module20_46_opt, self.reshape_1183_shape)
+ opt_split_1184, opt_split_1184_1 = self.split_1184(opt_reshape_1183)
+ opt_mul_1185 = P.Mul()(opt_split_1184, opt_split_1171)
+ opt_mul_1186 = P.Mul()(opt_split_1184_1, opt_split_1171_1)
+ opt_add_1187 = opt_mul_1185 + self.add_1187_bias
+ opt_add_1188 = P.Add()(opt_add_1187, opt_mul_1186)
+ opt_conv2d_1189 = self.conv2d_1189(opt_add_1188)
+ opt_add_1190 = P.Add()(opt_conv2d_1189, opt_relu_1166)
+ opt_relu_1191 = self.relu_1191(opt_add_1190)
+ module22_48_opt = self.module22_48(opt_relu_1191)
+ opt_split_1196, opt_split_1196_1 = self.split_1196(module22_48_opt)
+ opt_add_1197 = opt_split_1196 + self.add_1197_bias
+ opt_add_1198 = P.Add()(opt_add_1197, opt_split_1196_1)
+ module20_47_opt = self.module20_47(opt_add_1198)
+ opt_reshape_1208 = self.reshape_1208(module20_47_opt, self.reshape_1208_shape)
+ opt_split_1209, opt_split_1209_1 = self.split_1209(opt_reshape_1208)
+ opt_mul_1210 = P.Mul()(opt_split_1209, opt_split_1196)
+ opt_mul_1211 = P.Mul()(opt_split_1209_1, opt_split_1196_1)
+ opt_add_1212 = opt_mul_1210 + self.add_1212_bias
+ opt_add_1213 = P.Add()(opt_add_1212, opt_mul_1211)
+ opt_conv2d_1214 = self.conv2d_1214(opt_add_1213)
+ opt_add_1215 = P.Add()(opt_conv2d_1214, opt_relu_1191)
+ opt_relu_1216 = self.relu_1216(opt_add_1215)
+ module22_49_opt = self.module22_49(opt_relu_1216)
+ opt_split_1221, opt_split_1221_1 = self.split_1221(module22_49_opt)
+ opt_add_1222 = opt_split_1221 + self.add_1222_bias
+ opt_add_1223 = P.Add()(opt_add_1222, opt_split_1221_1)
+ module20_48_opt = self.module20_48(opt_add_1223)
+ opt_reshape_1233 = self.reshape_1233(module20_48_opt, self.reshape_1233_shape)
+ opt_split_1234, opt_split_1234_1 = self.split_1234(opt_reshape_1233)
+ opt_mul_1235 = P.Mul()(opt_split_1234, opt_split_1221)
+ opt_mul_1236 = P.Mul()(opt_split_1234_1, opt_split_1221_1)
+ opt_add_1237 = opt_mul_1235 + self.add_1237_bias
+ opt_add_1238 = P.Add()(opt_add_1237, opt_mul_1236)
+ opt_conv2d_1239 = self.conv2d_1239(opt_add_1238)
+ opt_add_1240 = P.Add()(opt_conv2d_1239, opt_relu_1216)
+ opt_relu_1241 = self.relu_1241(opt_add_1240)
+ module22_50_opt = self.module22_50(opt_relu_1241)
+ opt_split_1246, opt_split_1246_1 = self.split_1246(module22_50_opt)
+ opt_add_1247 = opt_split_1246 + self.add_1247_bias
+ opt_add_1248 = P.Add()(opt_add_1247, opt_split_1246_1)
+ module20_49_opt = self.module20_49(opt_add_1248)
+ opt_reshape_1258 = self.reshape_1258(module20_49_opt, self.reshape_1258_shape)
+ opt_split_1259, opt_split_1259_1 = self.split_1259(opt_reshape_1258)
+ opt_mul_1260 = P.Mul()(opt_split_1259, opt_split_1246)
+ opt_mul_1261 = P.Mul()(opt_split_1259_1, opt_split_1246_1)
+ opt_add_1262 = opt_mul_1260 + self.add_1262_bias
+ opt_add_1263 = P.Add()(opt_add_1262, opt_mul_1261)
+ opt_conv2d_1264 = self.conv2d_1264(opt_add_1263)
+ opt_add_1265 = P.Add()(opt_conv2d_1264, opt_relu_1241)
+ opt_relu_1266 = self.relu_1266(opt_add_1265)
+ module22_51_opt = self.module22_51(opt_relu_1266)
+ opt_split_1271, opt_split_1271_1 = self.split_1271(module22_51_opt)
+ opt_add_1272 = opt_split_1271 + self.add_1272_bias
+ opt_add_1273 = P.Add()(opt_add_1272, opt_split_1271_1)
+ module20_50_opt = self.module20_50(opt_add_1273)
+ opt_reshape_1283 = self.reshape_1283(module20_50_opt, self.reshape_1283_shape)
+ opt_split_1284, opt_split_1284_1 = self.split_1284(opt_reshape_1283)
+ opt_mul_1285 = P.Mul()(opt_split_1284, opt_split_1271)
+ opt_mul_1286 = P.Mul()(opt_split_1284_1, opt_split_1271_1)
+ opt_add_1287 = opt_mul_1285 + self.add_1287_bias
+ opt_add_1288 = P.Add()(opt_add_1287, opt_mul_1286)
+ opt_conv2d_1289 = self.conv2d_1289(opt_add_1288)
+ opt_add_1290 = P.Add()(opt_conv2d_1289, opt_relu_1266)
+ opt_relu_1291 = self.relu_1291(opt_add_1290)
+ module22_52_opt = self.module22_52(opt_relu_1291)
+ opt_split_1296, opt_split_1296_1 = self.split_1296(module22_52_opt)
+ opt_add_1297 = opt_split_1296 + self.add_1297_bias
+ opt_add_1298 = P.Add()(opt_add_1297, opt_split_1296_1)
+ module20_51_opt = self.module20_51(opt_add_1298)
+ opt_reshape_1308 = self.reshape_1308(module20_51_opt, self.reshape_1308_shape)
+ opt_split_1309, opt_split_1309_1 = self.split_1309(opt_reshape_1308)
+ opt_mul_1310 = P.Mul()(opt_split_1309, opt_split_1296)
+ opt_mul_1311 = P.Mul()(opt_split_1309_1, opt_split_1296_1)
+ opt_add_1312 = opt_mul_1310 + self.add_1312_bias
+ opt_add_1313 = P.Add()(opt_add_1312, opt_mul_1311)
+ opt_conv2d_1314 = self.conv2d_1314(opt_add_1313)
+ opt_add_1315 = P.Add()(opt_conv2d_1314, opt_relu_1291)
+ opt_relu_1316 = self.relu_1316(opt_add_1315)
+ module22_53_opt = self.module22_53(opt_relu_1316)
+ opt_split_1321, opt_split_1321_1 = self.split_1321(module22_53_opt)
+ opt_add_1322 = opt_split_1321 + self.add_1322_bias
+ opt_add_1323 = P.Add()(opt_add_1322, opt_split_1321_1)
+ module20_52_opt = self.module20_52(opt_add_1323)
+ opt_reshape_1333 = self.reshape_1333(module20_52_opt, self.reshape_1333_shape)
+ opt_split_1334, opt_split_1334_1 = self.split_1334(opt_reshape_1333)
+ opt_mul_1335 = P.Mul()(opt_split_1334, opt_split_1321)
+ opt_mul_1336 = P.Mul()(opt_split_1334_1, opt_split_1321_1)
+ opt_add_1337 = opt_mul_1335 + self.add_1337_bias
+ opt_add_1338 = P.Add()(opt_add_1337, opt_mul_1336)
+ opt_conv2d_1339 = self.conv2d_1339(opt_add_1338)
+ opt_add_1340 = P.Add()(opt_conv2d_1339, opt_relu_1316)
+ opt_relu_1341 = self.relu_1341(opt_add_1340)
+ module22_54_opt = self.module22_54(opt_relu_1341)
+ opt_split_1346, opt_split_1346_1 = self.split_1346(module22_54_opt)
+ opt_add_1347 = opt_split_1346 + self.add_1347_bias
+ opt_add_1348 = P.Add()(opt_add_1347, opt_split_1346_1)
+ module20_53_opt = self.module20_53(opt_add_1348)
+ opt_reshape_1358 = self.reshape_1358(module20_53_opt, self.reshape_1358_shape)
+ opt_split_1359, opt_split_1359_1 = self.split_1359(opt_reshape_1358)
+ opt_mul_1360 = P.Mul()(opt_split_1359, opt_split_1346)
+ opt_mul_1361 = P.Mul()(opt_split_1359_1, opt_split_1346_1)
+ opt_add_1362 = opt_mul_1360 + self.add_1362_bias
+ opt_add_1363 = P.Add()(opt_add_1362, opt_mul_1361)
+ opt_conv2d_1364 = self.conv2d_1364(opt_add_1363)
+ opt_add_1365 = P.Add()(opt_conv2d_1364, opt_relu_1341)
+ opt_relu_1366 = self.relu_1366(opt_add_1365)
+ module22_55_opt = self.module22_55(opt_relu_1366)
+ opt_split_1371, opt_split_1371_1 = self.split_1371(module22_55_opt)
+ opt_add_1372 = opt_split_1371 + self.add_1372_bias
+ opt_add_1373 = P.Add()(opt_add_1372, opt_split_1371_1)
+ module20_54_opt = self.module20_54(opt_add_1373)
+ opt_reshape_1383 = self.reshape_1383(module20_54_opt, self.reshape_1383_shape)
+ opt_split_1384, opt_split_1384_1 = self.split_1384(opt_reshape_1383)
+ opt_mul_1385 = P.Mul()(opt_split_1384, opt_split_1371)
+ opt_mul_1386 = P.Mul()(opt_split_1384_1, opt_split_1371_1)
+ opt_add_1387 = opt_mul_1385 + self.add_1387_bias
+ opt_add_1388 = P.Add()(opt_add_1387, opt_mul_1386)
+ opt_conv2d_1389 = self.conv2d_1389(opt_add_1388)
+ opt_add_1390 = P.Add()(opt_conv2d_1389, opt_relu_1366)
+ opt_relu_1391 = self.relu_1391(opt_add_1390)
+ module22_56_opt = self.module22_56(opt_relu_1391)
+ opt_split_1396, opt_split_1396_1 = self.split_1396(module22_56_opt)
+ opt_add_1397 = opt_split_1396 + self.add_1397_bias
+ opt_add_1398 = P.Add()(opt_add_1397, opt_split_1396_1)
+ module20_55_opt = self.module20_55(opt_add_1398)
+ opt_reshape_1408 = self.reshape_1408(module20_55_opt, self.reshape_1408_shape)
+ opt_split_1409, opt_split_1409_1 = self.split_1409(opt_reshape_1408)
+ opt_mul_1410 = P.Mul()(opt_split_1409, opt_split_1396)
+ opt_mul_1411 = P.Mul()(opt_split_1409_1, opt_split_1396_1)
+ opt_add_1412 = opt_mul_1410 + self.add_1412_bias
+ opt_add_1413 = P.Add()(opt_add_1412, opt_mul_1411)
+ opt_conv2d_1414 = self.conv2d_1414(opt_add_1413)
+ opt_add_1415 = P.Add()(opt_conv2d_1414, opt_relu_1391)
+ opt_relu_1416 = self.relu_1416(opt_add_1415)
+ module22_57_opt = self.module22_57(opt_relu_1416)
+ opt_split_1421, opt_split_1421_1 = self.split_1421(module22_57_opt)
+ opt_add_1422 = opt_split_1421 + self.add_1422_bias
+ opt_add_1423 = P.Add()(opt_add_1422, opt_split_1421_1)
+ module20_56_opt = self.module20_56(opt_add_1423)
+ opt_reshape_1433 = self.reshape_1433(module20_56_opt, self.reshape_1433_shape)
+ opt_split_1434, opt_split_1434_1 = self.split_1434(opt_reshape_1433)
+ opt_mul_1435 = P.Mul()(opt_split_1434, opt_split_1421)
+ opt_mul_1436 = P.Mul()(opt_split_1434_1, opt_split_1421_1)
+ opt_add_1437 = opt_mul_1435 + self.add_1437_bias
+ opt_add_1438 = P.Add()(opt_add_1437, opt_mul_1436)
+ opt_conv2d_1439 = self.conv2d_1439(opt_add_1438)
+ opt_add_1440 = P.Add()(opt_conv2d_1439, opt_relu_1416)
+ opt_relu_1441 = self.relu_1441(opt_add_1440)
+ module22_58_opt = self.module22_58(opt_relu_1441)
+ opt_split_1446, opt_split_1446_1 = self.split_1446(module22_58_opt)
+ opt_add_1447 = opt_split_1446 + self.add_1447_bias
+ opt_add_1448 = P.Add()(opt_add_1447, opt_split_1446_1)
+ module20_57_opt = self.module20_57(opt_add_1448)
+ opt_reshape_1458 = self.reshape_1458(module20_57_opt, self.reshape_1458_shape)
+ opt_split_1459, opt_split_1459_1 = self.split_1459(opt_reshape_1458)
+ opt_mul_1460 = P.Mul()(opt_split_1459, opt_split_1446)
+ opt_mul_1461 = P.Mul()(opt_split_1459_1, opt_split_1446_1)
+ opt_add_1462 = opt_mul_1460 + self.add_1462_bias
+ opt_add_1463 = P.Add()(opt_add_1462, opt_mul_1461)
+ opt_conv2d_1464 = self.conv2d_1464(opt_add_1463)
+ opt_add_1465 = P.Add()(opt_conv2d_1464, opt_relu_1441)
+ opt_relu_1466 = self.relu_1466(opt_add_1465)
+ module22_59_opt = self.module22_59(opt_relu_1466)
+ opt_split_1471, opt_split_1471_1 = self.split_1471(module22_59_opt)
+ opt_add_1472 = opt_split_1471 + self.add_1472_bias
+ opt_add_1473 = P.Add()(opt_add_1472, opt_split_1471_1)
+ module20_58_opt = self.module20_58(opt_add_1473)
+ opt_reshape_1483 = self.reshape_1483(module20_58_opt, self.reshape_1483_shape)
+ opt_split_1484, opt_split_1484_1 = self.split_1484(opt_reshape_1483)
+ opt_mul_1485 = P.Mul()(opt_split_1484, opt_split_1471)
+ opt_mul_1486 = P.Mul()(opt_split_1484_1, opt_split_1471_1)
+ opt_add_1487 = opt_mul_1485 + self.add_1487_bias
+ opt_add_1488 = P.Add()(opt_add_1487, opt_mul_1486)
+ opt_conv2d_1489 = self.conv2d_1489(opt_add_1488)
+ opt_add_1490 = P.Add()(opt_conv2d_1489, opt_relu_1466)
+ opt_relu_1491 = self.relu_1491(opt_add_1490)
+ module22_60_opt = self.module22_60(opt_relu_1491)
+ opt_split_1496, opt_split_1496_1 = self.split_1496(module22_60_opt)
+ opt_add_1497 = opt_split_1496 + self.add_1497_bias
+ opt_add_1498 = P.Add()(opt_add_1497, opt_split_1496_1)
+ module20_59_opt = self.module20_59(opt_add_1498)
+ opt_reshape_1508 = self.reshape_1508(module20_59_opt, self.reshape_1508_shape)
+ opt_split_1509, opt_split_1509_1 = self.split_1509(opt_reshape_1508)
+ opt_mul_1510 = P.Mul()(opt_split_1509, opt_split_1496)
+ opt_mul_1511 = P.Mul()(opt_split_1509_1, opt_split_1496_1)
+ opt_add_1512 = opt_mul_1510 + self.add_1512_bias
+ opt_add_1513 = P.Add()(opt_add_1512, opt_mul_1511)
+ opt_conv2d_1514 = self.conv2d_1514(opt_add_1513)
+ opt_add_1515 = P.Add()(opt_conv2d_1514, opt_relu_1491)
+ opt_relu_1516 = self.relu_1516(opt_add_1515)
+ module22_61_opt = self.module22_61(opt_relu_1516)
+ opt_split_1521, opt_split_1521_1 = self.split_1521(module22_61_opt)
+ opt_add_1522 = opt_split_1521 + self.add_1522_bias
+ opt_add_1523 = P.Add()(opt_add_1522, opt_split_1521_1)
+ module20_60_opt = self.module20_60(opt_add_1523)
+ opt_reshape_1533 = self.reshape_1533(module20_60_opt, self.reshape_1533_shape)
+ opt_split_1534, opt_split_1534_1 = self.split_1534(opt_reshape_1533)
+ opt_mul_1535 = P.Mul()(opt_split_1534, opt_split_1521)
+ opt_mul_1536 = P.Mul()(opt_split_1534_1, opt_split_1521_1)
+ opt_add_1537 = opt_mul_1535 + self.add_1537_bias
+ opt_add_1538 = P.Add()(opt_add_1537, opt_mul_1536)
+ opt_conv2d_1539 = self.conv2d_1539(opt_add_1538)
+ opt_add_1540 = P.Add()(opt_conv2d_1539, opt_relu_1516)
+ opt_relu_1541 = self.relu_1541(opt_add_1540)
+ module22_62_opt = self.module22_62(opt_relu_1541)
+ opt_split_1546, opt_split_1546_1 = self.split_1546(module22_62_opt)
+ opt_add_1547 = opt_split_1546 + self.add_1547_bias
+ opt_add_1548 = P.Add()(opt_add_1547, opt_split_1546_1)
+ module20_61_opt = self.module20_61(opt_add_1548)
+ opt_reshape_1558 = self.reshape_1558(module20_61_opt, self.reshape_1558_shape)
+ opt_split_1559, opt_split_1559_1 = self.split_1559(opt_reshape_1558)
+ opt_mul_1560 = P.Mul()(opt_split_1559, opt_split_1546)
+ opt_mul_1561 = P.Mul()(opt_split_1559_1, opt_split_1546_1)
+ opt_add_1562 = opt_mul_1560 + self.add_1562_bias
+ opt_add_1563 = P.Add()(opt_add_1562, opt_mul_1561)
+ opt_conv2d_1564 = self.conv2d_1564(opt_add_1563)
+ opt_add_1565 = P.Add()(opt_conv2d_1564, opt_relu_1541)
+ opt_relu_1566 = self.relu_1566(opt_add_1565)
+ module22_63_opt = self.module22_63(opt_relu_1566)
+ opt_split_1571, opt_split_1571_1 = self.split_1571(module22_63_opt)
+ opt_add_1572 = opt_split_1571 + self.add_1572_bias
+ opt_add_1573 = P.Add()(opt_add_1572, opt_split_1571_1)
+ module20_62_opt = self.module20_62(opt_add_1573)
+ opt_reshape_1583 = self.reshape_1583(module20_62_opt, self.reshape_1583_shape)
+ opt_split_1584, opt_split_1584_1 = self.split_1584(opt_reshape_1583)
+ opt_mul_1585 = P.Mul()(opt_split_1584, opt_split_1571)
+ opt_mul_1586 = P.Mul()(opt_split_1584_1, opt_split_1571_1)
+ opt_add_1587 = opt_mul_1585 + self.add_1587_bias
+ opt_add_1588 = P.Add()(opt_add_1587, opt_mul_1586)
+ opt_conv2d_1589 = self.conv2d_1589(opt_add_1588)
+ opt_add_1590 = P.Add()(opt_conv2d_1589, opt_relu_1566)
+ opt_relu_1591 = self.relu_1591(opt_add_1590)
+ module22_64_opt = self.module22_64(opt_relu_1591)
+ opt_split_1596, opt_split_1596_1 = self.split_1596(module22_64_opt)
+ opt_add_1597 = opt_split_1596 + self.add_1597_bias
+ opt_add_1598 = P.Add()(opt_add_1597, opt_split_1596_1)
+ module20_63_opt = self.module20_63(opt_add_1598)
+ opt_reshape_1608 = self.reshape_1608(module20_63_opt, self.reshape_1608_shape)
+ opt_split_1609, opt_split_1609_1 = self.split_1609(opt_reshape_1608)
+ opt_mul_1610 = P.Mul()(opt_split_1609, opt_split_1596)
+ opt_mul_1611 = P.Mul()(opt_split_1609_1, opt_split_1596_1)
+ opt_add_1612 = opt_mul_1610 + self.add_1612_bias
+ opt_add_1613 = P.Add()(opt_add_1612, opt_mul_1611)
+ opt_conv2d_1614 = self.conv2d_1614(opt_add_1613)
+ opt_add_1615 = P.Add()(opt_conv2d_1614, opt_relu_1591)
+ opt_relu_1616 = self.relu_1616(opt_add_1615)
+ module22_65_opt = self.module22_65(opt_relu_1616)
+ opt_split_1621, opt_split_1621_1 = self.split_1621(module22_65_opt)
+ opt_add_1622 = opt_split_1621 + self.add_1622_bias
+ opt_add_1623 = P.Add()(opt_add_1622, opt_split_1621_1)
+ module20_64_opt = self.module20_64(opt_add_1623)
+ opt_reshape_1633 = self.reshape_1633(module20_64_opt, self.reshape_1633_shape)
+ opt_split_1634, opt_split_1634_1 = self.split_1634(opt_reshape_1633)
+ opt_mul_1635 = P.Mul()(opt_split_1634, opt_split_1621)
+ opt_mul_1636 = P.Mul()(opt_split_1634_1, opt_split_1621_1)
+ opt_add_1637 = opt_mul_1635 + self.add_1637_bias
+ opt_add_1638 = P.Add()(opt_add_1637, opt_mul_1636)
+ opt_conv2d_1639 = self.conv2d_1639(opt_add_1638)
+ opt_add_1640 = P.Add()(opt_conv2d_1639, opt_relu_1616)
+ opt_relu_1641 = self.relu_1641(opt_add_1640)
+ module22_66_opt = self.module22_66(opt_relu_1641)
+ opt_split_1646, opt_split_1646_1 = self.split_1646(module22_66_opt)
+ opt_add_1647 = opt_split_1646 + self.add_1647_bias
+ opt_add_1648 = P.Add()(opt_add_1647, opt_split_1646_1)
+ module20_65_opt = self.module20_65(opt_add_1648)
+ opt_reshape_1658 = self.reshape_1658(module20_65_opt, self.reshape_1658_shape)
+ opt_split_1659, opt_split_1659_1 = self.split_1659(opt_reshape_1658)
+ opt_mul_1660 = P.Mul()(opt_split_1659, opt_split_1646)
+ opt_mul_1661 = P.Mul()(opt_split_1659_1, opt_split_1646_1)
+ opt_add_1662 = opt_mul_1660 + self.add_1662_bias
+ opt_add_1663 = P.Add()(opt_add_1662, opt_mul_1661)
+ opt_conv2d_1664 = self.conv2d_1664(opt_add_1663)
+ opt_add_1665 = P.Add()(opt_conv2d_1664, opt_relu_1641)
+ opt_relu_1666 = self.relu_1666(opt_add_1665)
+ module22_67_opt = self.module22_67(opt_relu_1666)
+ opt_split_1671, opt_split_1671_1 = self.split_1671(module22_67_opt)
+ opt_add_1672 = opt_split_1671 + self.add_1672_bias
+ opt_add_1673 = P.Add()(opt_add_1672, opt_split_1671_1)
+ module20_66_opt = self.module20_66(opt_add_1673)
+ opt_reshape_1683 = self.reshape_1683(module20_66_opt, self.reshape_1683_shape)
+ opt_split_1684, opt_split_1684_1 = self.split_1684(opt_reshape_1683)
+ opt_mul_1685 = P.Mul()(opt_split_1684, opt_split_1671)
+ opt_mul_1686 = P.Mul()(opt_split_1684_1, opt_split_1671_1)
+ opt_add_1687 = opt_mul_1685 + self.add_1687_bias
+ opt_add_1688 = P.Add()(opt_add_1687, opt_mul_1686)
+ opt_conv2d_1689 = self.conv2d_1689(opt_add_1688)
+ opt_add_1690 = P.Add()(opt_conv2d_1689, opt_relu_1666)
+ opt_relu_1691 = self.relu_1691(opt_add_1690)
+ module22_68_opt = self.module22_68(opt_relu_1691)
+ opt_split_1696, opt_split_1696_1 = self.split_1696(module22_68_opt)
+ opt_add_1697 = opt_split_1696 + self.add_1697_bias
+ opt_add_1698 = P.Add()(opt_add_1697, opt_split_1696_1)
+ module20_67_opt = self.module20_67(opt_add_1698)
+ opt_reshape_1708 = self.reshape_1708(module20_67_opt, self.reshape_1708_shape)
+ opt_split_1709, opt_split_1709_1 = self.split_1709(opt_reshape_1708)
+ opt_mul_1710 = P.Mul()(opt_split_1709, opt_split_1696)
+ opt_mul_1711 = P.Mul()(opt_split_1709_1, opt_split_1696_1)
+ opt_add_1712 = opt_mul_1710 + self.add_1712_bias
+ opt_add_1713 = P.Add()(opt_add_1712, opt_mul_1711)
+ opt_conv2d_1714 = self.conv2d_1714(opt_add_1713)
+ opt_add_1715 = P.Add()(opt_conv2d_1714, opt_relu_1691)
+ opt_relu_1716 = self.relu_1716(opt_add_1715)
+ module22_69_opt = self.module22_69(opt_relu_1716)
+ opt_split_1721, opt_split_1721_1 = self.split_1721(module22_69_opt)
+ opt_add_1722 = opt_split_1721 + self.add_1722_bias
+ opt_add_1723 = P.Add()(opt_add_1722, opt_split_1721_1)
+ module20_68_opt = self.module20_68(opt_add_1723)
+ opt_reshape_1733 = self.reshape_1733(module20_68_opt, self.reshape_1733_shape)
+ opt_split_1734, opt_split_1734_1 = self.split_1734(opt_reshape_1733)
+ opt_mul_1735 = P.Mul()(opt_split_1734, opt_split_1721)
+ opt_mul_1736 = P.Mul()(opt_split_1734_1, opt_split_1721_1)
+ opt_add_1737 = opt_mul_1735 + self.add_1737_bias
+ opt_add_1738 = P.Add()(opt_add_1737, opt_mul_1736)
+ opt_conv2d_1739 = self.conv2d_1739(opt_add_1738)
+ opt_add_1740 = P.Add()(opt_conv2d_1739, opt_relu_1716)
+ opt_relu_1741 = self.relu_1741(opt_add_1740)
+ module22_70_opt = self.module22_70(opt_relu_1741)
+ opt_split_1746, opt_split_1746_1 = self.split_1746(module22_70_opt)
+ opt_add_1747 = opt_split_1746 + self.add_1747_bias
+ opt_add_1748 = P.Add()(opt_add_1747, opt_split_1746_1)
+ module20_69_opt = self.module20_69(opt_add_1748)
+ opt_reshape_1758 = self.reshape_1758(module20_69_opt, self.reshape_1758_shape)
+ opt_split_1759, opt_split_1759_1 = self.split_1759(opt_reshape_1758)
+ opt_mul_1760 = P.Mul()(opt_split_1759, opt_split_1746)
+ opt_mul_1761 = P.Mul()(opt_split_1759_1, opt_split_1746_1)
+ opt_add_1762 = opt_mul_1760 + self.add_1762_bias
+ opt_add_1763 = P.Add()(opt_add_1762, opt_mul_1761)
+ opt_conv2d_1764 = self.conv2d_1764(opt_add_1763)
+ opt_add_1765 = P.Add()(opt_conv2d_1764, opt_relu_1741)
+ opt_relu_1766 = self.relu_1766(opt_add_1765)
+ module22_71_opt = self.module22_71(opt_relu_1766)
+ opt_split_1771, opt_split_1771_1 = self.split_1771(module22_71_opt)
+ opt_add_1772 = opt_split_1771 + self.add_1772_bias
+ opt_add_1773 = P.Add()(opt_add_1772, opt_split_1771_1)
+ module20_70_opt = self.module20_70(opt_add_1773)
+ opt_reshape_1783 = self.reshape_1783(module20_70_opt, self.reshape_1783_shape)
+ opt_split_1784, opt_split_1784_1 = self.split_1784(opt_reshape_1783)
+ opt_mul_1785 = P.Mul()(opt_split_1784, opt_split_1771)
+ opt_mul_1786 = P.Mul()(opt_split_1784_1, opt_split_1771_1)
+ opt_add_1787 = opt_mul_1785 + self.add_1787_bias
+ opt_add_1788 = P.Add()(opt_add_1787, opt_mul_1786)
+ opt_conv2d_1789 = self.conv2d_1789(opt_add_1788)
+ opt_add_1790 = P.Add()(opt_conv2d_1789, opt_relu_1766)
+ opt_relu_1791 = self.relu_1791(opt_add_1790)
+ module22_72_opt = self.module22_72(opt_relu_1791)
+ opt_split_1796, opt_split_1796_1 = self.split_1796(module22_72_opt)
+ opt_add_1797 = opt_split_1796 + self.add_1797_bias
+ opt_add_1798 = P.Add()(opt_add_1797, opt_split_1796_1)
+ module20_71_opt = self.module20_71(opt_add_1798)
+ opt_reshape_1808 = self.reshape_1808(module20_71_opt, self.reshape_1808_shape)
+ opt_split_1809, opt_split_1809_1 = self.split_1809(opt_reshape_1808)
+ opt_mul_1810 = P.Mul()(opt_split_1809, opt_split_1796)
+ opt_mul_1811 = P.Mul()(opt_split_1809_1, opt_split_1796_1)
+ opt_add_1812 = opt_mul_1810 + self.add_1812_bias
+ opt_add_1813 = P.Add()(opt_add_1812, opt_mul_1811)
+ opt_conv2d_1814 = self.conv2d_1814(opt_add_1813)
+ opt_add_1815 = P.Add()(opt_conv2d_1814, opt_relu_1791)
+ opt_relu_1816 = self.relu_1816(opt_add_1815)
+ module22_73_opt = self.module22_73(opt_relu_1816)
+ opt_split_1821, opt_split_1821_1 = self.split_1821(module22_73_opt)
+ opt_add_1822 = opt_split_1821 + self.add_1822_bias
+ opt_add_1823 = P.Add()(opt_add_1822, opt_split_1821_1)
+ module20_72_opt = self.module20_72(opt_add_1823)
+ opt_reshape_1833 = self.reshape_1833(module20_72_opt, self.reshape_1833_shape)
+ opt_split_1834, opt_split_1834_1 = self.split_1834(opt_reshape_1833)
+ opt_mul_1835 = P.Mul()(opt_split_1834, opt_split_1821)
+ opt_mul_1836 = P.Mul()(opt_split_1834_1, opt_split_1821_1)
+ opt_add_1837 = opt_mul_1835 + self.add_1837_bias
+ opt_add_1838 = P.Add()(opt_add_1837, opt_mul_1836)
+ opt_conv2d_1839 = self.conv2d_1839(opt_add_1838)
+ opt_add_1840 = P.Add()(opt_conv2d_1839, opt_relu_1816)
+ opt_relu_1841 = self.relu_1841(opt_add_1840)
+ module22_74_opt = self.module22_74(opt_relu_1841)
+ opt_split_1846, opt_split_1846_1 = self.split_1846(module22_74_opt)
+ opt_add_1847 = opt_split_1846 + self.add_1847_bias
+ opt_add_1848 = P.Add()(opt_add_1847, opt_split_1846_1)
+ module20_73_opt = self.module20_73(opt_add_1848)
+ opt_reshape_1858 = self.reshape_1858(module20_73_opt, self.reshape_1858_shape)
+ opt_split_1859, opt_split_1859_1 = self.split_1859(opt_reshape_1858)
+ opt_mul_1860 = P.Mul()(opt_split_1859, opt_split_1846)
+ opt_mul_1861 = P.Mul()(opt_split_1859_1, opt_split_1846_1)
+ opt_add_1862 = opt_mul_1860 + self.add_1862_bias
+ opt_add_1863 = P.Add()(opt_add_1862, opt_mul_1861)
+ opt_conv2d_1864 = self.conv2d_1864(opt_add_1863)
+ opt_add_1865 = P.Add()(opt_conv2d_1864, opt_relu_1841)
+ opt_relu_1866 = self.relu_1866(opt_add_1865)
+ module22_75_opt = self.module22_75(opt_relu_1866)
+ opt_split_1871, opt_split_1871_1 = self.split_1871(module22_75_opt)
+ opt_add_1872 = opt_split_1871 + self.add_1872_bias
+ opt_add_1873 = P.Add()(opt_add_1872, opt_split_1871_1)
+ module20_74_opt = self.module20_74(opt_add_1873)
+ opt_reshape_1883 = self.reshape_1883(module20_74_opt, self.reshape_1883_shape)
+ opt_split_1884, opt_split_1884_1 = self.split_1884(opt_reshape_1883)
+ opt_mul_1885 = P.Mul()(opt_split_1884, opt_split_1871)
+ opt_mul_1886 = P.Mul()(opt_split_1884_1, opt_split_1871_1)
+ opt_add_1887 = opt_mul_1885 + self.add_1887_bias
+ opt_add_1888 = P.Add()(opt_add_1887, opt_mul_1886)
+ opt_conv2d_1889 = self.conv2d_1889(opt_add_1888)
+ opt_add_1890 = P.Add()(opt_conv2d_1889, opt_relu_1866)
+ opt_relu_1891 = self.relu_1891(opt_add_1890)
+ module22_76_opt = self.module22_76(opt_relu_1891)
+ opt_split_1896, opt_split_1896_1 = self.split_1896(module22_76_opt)
+ opt_add_1897 = opt_split_1896 + self.add_1897_bias
+ opt_add_1898 = P.Add()(opt_add_1897, opt_split_1896_1)
+ module20_75_opt = self.module20_75(opt_add_1898)
+ opt_reshape_1908 = self.reshape_1908(module20_75_opt, self.reshape_1908_shape)
+ opt_split_1909, opt_split_1909_1 = self.split_1909(opt_reshape_1908)
+ opt_mul_1910 = P.Mul()(opt_split_1909, opt_split_1896)
+ opt_mul_1911 = P.Mul()(opt_split_1909_1, opt_split_1896_1)
+ opt_add_1912 = opt_mul_1910 + self.add_1912_bias
+ opt_add_1913 = P.Add()(opt_add_1912, opt_mul_1911)
+ opt_conv2d_1914 = self.conv2d_1914(opt_add_1913)
+ opt_add_1915 = P.Add()(opt_conv2d_1914, opt_relu_1891)
+ opt_relu_1916 = self.relu_1916(opt_add_1915)
+ module22_77_opt = self.module22_77(opt_relu_1916)
+ opt_split_1921, opt_split_1921_1 = self.split_1921(module22_77_opt)
+ opt_add_1922 = opt_split_1921 + self.add_1922_bias
+ opt_add_1923 = P.Add()(opt_add_1922, opt_split_1921_1)
+ module20_76_opt = self.module20_76(opt_add_1923)
+ opt_reshape_1933 = self.reshape_1933(module20_76_opt, self.reshape_1933_shape)
+ opt_split_1934, opt_split_1934_1 = self.split_1934(opt_reshape_1933)
+ opt_mul_1935 = P.Mul()(opt_split_1934, opt_split_1921)
+ opt_mul_1936 = P.Mul()(opt_split_1934_1, opt_split_1921_1)
+ opt_add_1937 = opt_mul_1935 + self.add_1937_bias
+ opt_add_1938 = P.Add()(opt_add_1937, opt_mul_1936)
+ opt_conv2d_1939 = self.conv2d_1939(opt_add_1938)
+ opt_add_1940 = P.Add()(opt_conv2d_1939, opt_relu_1916)
+ opt_relu_1941 = self.relu_1941(opt_add_1940)
+ module22_78_opt = self.module22_78(opt_relu_1941)
+ opt_split_1946, opt_split_1946_1 = self.split_1946(module22_78_opt)
+ opt_add_1947 = opt_split_1946 + self.add_1947_bias
+ opt_add_1948 = P.Add()(opt_add_1947, opt_split_1946_1)
+ module20_77_opt = self.module20_77(opt_add_1948)
+ opt_reshape_1958 = self.reshape_1958(module20_77_opt, self.reshape_1958_shape)
+ opt_split_1959, opt_split_1959_1 = self.split_1959(opt_reshape_1958)
+ opt_mul_1960 = P.Mul()(opt_split_1959, opt_split_1946)
+ opt_mul_1961 = P.Mul()(opt_split_1959_1, opt_split_1946_1)
+ opt_add_1962 = opt_mul_1960 + self.add_1962_bias
+ opt_add_1963 = P.Add()(opt_add_1962, opt_mul_1961)
+ opt_conv2d_1964 = self.conv2d_1964(opt_add_1963)
+ opt_add_1965 = P.Add()(opt_conv2d_1964, opt_relu_1941)
+ opt_relu_1966 = self.relu_1966(opt_add_1965)
+ module22_79_opt = self.module22_79(opt_relu_1966)
+ opt_split_1971, opt_split_1971_1 = self.split_1971(module22_79_opt)
+ opt_add_1972 = opt_split_1971 + self.add_1972_bias
+ opt_add_1973 = P.Add()(opt_add_1972, opt_split_1971_1)
+ module20_78_opt = self.module20_78(opt_add_1973)
+ opt_reshape_1983 = self.reshape_1983(module20_78_opt, self.reshape_1983_shape)
+ opt_split_1984, opt_split_1984_1 = self.split_1984(opt_reshape_1983)
+ opt_mul_1985 = P.Mul()(opt_split_1984, opt_split_1971)
+ opt_mul_1986 = P.Mul()(opt_split_1984_1, opt_split_1971_1)
+ opt_add_1987 = opt_mul_1985 + self.add_1987_bias
+ opt_add_1988 = P.Add()(opt_add_1987, opt_mul_1986)
+ opt_conv2d_1989 = self.conv2d_1989(opt_add_1988)
+ opt_add_1990 = P.Add()(opt_conv2d_1989, opt_relu_1966)
+ opt_relu_1991 = self.relu_1991(opt_add_1990)
+ module22_80_opt = self.module22_80(opt_relu_1991)
+ opt_split_1996, opt_split_1996_1 = self.split_1996(module22_80_opt)
+ opt_add_1997 = opt_split_1996 + self.add_1997_bias
+ opt_add_1998 = P.Add()(opt_add_1997, opt_split_1996_1)
+ module20_79_opt = self.module20_79(opt_add_1998)
+ opt_reshape_2008 = self.reshape_2008(module20_79_opt, self.reshape_2008_shape)
+ opt_split_2009, opt_split_2009_1 = self.split_2009(opt_reshape_2008)
+ opt_mul_2010 = P.Mul()(opt_split_2009, opt_split_1996)
+ opt_mul_2011 = P.Mul()(opt_split_2009_1, opt_split_1996_1)
+ opt_add_2012 = opt_mul_2010 + self.add_2012_bias
+ opt_add_2013 = P.Add()(opt_add_2012, opt_mul_2011)
+ opt_conv2d_2014 = self.conv2d_2014(opt_add_2013)
+ opt_add_2015 = P.Add()(opt_conv2d_2014, opt_relu_1991)
+ opt_relu_2016 = self.relu_2016(opt_add_2015)
+ module22_81_opt = self.module22_81(opt_relu_2016)
+ opt_split_2021, opt_split_2021_1 = self.split_2021(module22_81_opt)
+ opt_add_2022 = opt_split_2021 + self.add_2022_bias
+ opt_add_2023 = P.Add()(opt_add_2022, opt_split_2021_1)
+ module20_80_opt = self.module20_80(opt_add_2023)
+ opt_reshape_2033 = self.reshape_2033(module20_80_opt, self.reshape_2033_shape)
+ opt_split_2034, opt_split_2034_1 = self.split_2034(opt_reshape_2033)
+ opt_mul_2035 = P.Mul()(opt_split_2034, opt_split_2021)
+ opt_mul_2036 = P.Mul()(opt_split_2034_1, opt_split_2021_1)
+ opt_add_2037 = opt_mul_2035 + self.add_2037_bias
+ opt_add_2038 = P.Add()(opt_add_2037, opt_mul_2036)
+ opt_conv2d_2039 = self.conv2d_2039(opt_add_2038)
+ opt_add_2040 = P.Add()(opt_conv2d_2039, opt_relu_2016)
+ opt_relu_2041 = self.relu_2041(opt_add_2040)
+ module22_82_opt = self.module22_82(opt_relu_2041)
+ opt_split_2048, opt_split_2048_1 = self.split_2048(module22_82_opt)
+ opt_add_2049 = opt_split_2048 + self.add_2049_bias
+ opt_add_2050 = P.Add()(opt_add_2049, opt_split_2048_1)
+ module20_81_opt = self.module20_81(opt_add_2050)
+ opt_reshape_2060 = self.reshape_2060(module20_81_opt, self.reshape_2060_shape)
+ opt_split_2061, opt_split_2061_1 = self.split_2061(opt_reshape_2060)
+ opt_mul_2062 = P.Mul()(opt_split_2061, opt_split_2048)
+ opt_mul_2063 = P.Mul()(opt_split_2061_1, opt_split_2048_1)
+ opt_add_2064 = opt_mul_2062 + self.add_2064_bias
+ opt_add_2065 = P.Add()(opt_add_2064, opt_mul_2063)
+ opt_pad_2066 = self.pad_2066(opt_add_2065)
+ opt_avgpool2d_2067 = self.pad_avgpool2d_2067(opt_pad_2066)
+ opt_avgpool2d_2067 = self.avgpool2d_2067(opt_avgpool2d_2067)
+ opt_conv2d_2068 = self.conv2d_2068(opt_avgpool2d_2067)
+ opt_avgpool2d_2043 = self.pad_avgpool2d_2043(opt_relu_2041)
+ opt_avgpool2d_2043 = self.avgpool2d_2043(opt_avgpool2d_2043)
+ opt_conv2d_2045 = self.conv2d_2045(opt_avgpool2d_2043)
+ opt_add_2069 = P.Add()(opt_conv2d_2068, opt_conv2d_2045)
+ opt_relu_2070 = self.relu_2070(opt_add_2069)
+ module22_83_opt = self.module22_83(opt_relu_2070)
+ opt_split_2075, opt_split_2075_1 = self.split_2075(module22_83_opt)
+ opt_add_2076 = opt_split_2075 + self.add_2076_bias
+ opt_add_2077 = P.Add()(opt_add_2076, opt_split_2075_1)
+ module20_82_opt = self.module20_82(opt_add_2077)
+ opt_reshape_2087 = self.reshape_2087(module20_82_opt, self.reshape_2087_shape)
+ opt_split_2088, opt_split_2088_1 = self.split_2088(opt_reshape_2087)
+ opt_mul_2089 = P.Mul()(opt_split_2088, opt_split_2075)
+ opt_mul_2090 = P.Mul()(opt_split_2088_1, opt_split_2075_1)
+ opt_add_2091 = opt_mul_2089 + self.add_2091_bias
+ opt_add_2092 = P.Add()(opt_add_2091, opt_mul_2090)
+ opt_conv2d_2093 = self.conv2d_2093(opt_add_2092)
+ opt_add_2094 = P.Add()(opt_conv2d_2093, opt_relu_2070)
+ opt_relu_2095 = self.relu_2095(opt_add_2094)
+ module22_84_opt = self.module22_84(opt_relu_2095)
+ opt_split_2100, opt_split_2100_1 = self.split_2100(module22_84_opt)
+ opt_add_2101 = opt_split_2100 + self.add_2101_bias
+ opt_add_2102 = P.Add()(opt_add_2101, opt_split_2100_1)
+ module20_83_opt = self.module20_83(opt_add_2102)
+ opt_reshape_2112 = self.reshape_2112(module20_83_opt, self.reshape_2112_shape)
+ opt_split_2113, opt_split_2113_1 = self.split_2113(opt_reshape_2112)
+ opt_mul_2114 = P.Mul()(opt_split_2113, opt_split_2100)
+ opt_mul_2115 = P.Mul()(opt_split_2113_1, opt_split_2100_1)
+ opt_add_2116 = opt_mul_2114 + self.add_2116_bias
+ opt_add_2117 = P.Add()(opt_add_2116, opt_mul_2115)
+ opt_conv2d_2118 = self.conv2d_2118(opt_add_2117)
+ opt_add_2119 = P.Add()(opt_conv2d_2118, opt_relu_2095)
+ opt_relu_2120 = self.relu_2120(opt_add_2119)
+ module22_85_opt = self.module22_85(opt_relu_2120)
+ opt_split_2125, opt_split_2125_1 = self.split_2125(module22_85_opt)
+ opt_add_2126 = opt_split_2125 + self.add_2126_bias
+ opt_add_2127 = P.Add()(opt_add_2126, opt_split_2125_1)
+ module20_84_opt = self.module20_84(opt_add_2127)
+ opt_reshape_2137 = self.reshape_2137(module20_84_opt, self.reshape_2137_shape)
+ opt_split_2138, opt_split_2138_1 = self.split_2138(opt_reshape_2137)
+ opt_mul_2139 = P.Mul()(opt_split_2138, opt_split_2125)
+ opt_mul_2140 = P.Mul()(opt_split_2138_1, opt_split_2125_1)
+ opt_add_2141 = opt_mul_2139 + self.add_2141_bias
+ opt_add_2142 = P.Add()(opt_add_2141, opt_mul_2140)
+ opt_conv2d_2143 = self.conv2d_2143(opt_add_2142)
+ opt_add_2144 = P.Add()(opt_conv2d_2143, opt_relu_2120)
+ opt_relu_2145 = self.relu_2145(opt_add_2144)
+ module22_86_opt = self.module22_86(opt_relu_2145)
+ opt_split_2150, opt_split_2150_1 = self.split_2150(module22_86_opt)
+ opt_add_2151 = opt_split_2150 + self.add_2151_bias
+ opt_add_2152 = P.Add()(opt_add_2151, opt_split_2150_1)
+ module20_85_opt = self.module20_85(opt_add_2152)
+ opt_reshape_2162 = self.reshape_2162(module20_85_opt, self.reshape_2162_shape)
+ opt_split_2163, opt_split_2163_1 = self.split_2163(opt_reshape_2162)
+ opt_mul_2164 = P.Mul()(opt_split_2163, opt_split_2150)
+ opt_mul_2165 = P.Mul()(opt_split_2163_1, opt_split_2150_1)
+ opt_add_2166 = opt_mul_2164 + self.add_2166_bias
+ opt_add_2167 = P.Add()(opt_add_2166, opt_mul_2165)
+ opt_conv2d_2168 = self.conv2d_2168(opt_add_2167)
+ opt_add_2169 = P.Add()(opt_conv2d_2168, opt_relu_2145)
+ opt_relu_2170 = self.relu_2170(opt_add_2169)
+ module22_87_opt = self.module22_87(opt_relu_2170)
+ opt_split_2175, opt_split_2175_1 = self.split_2175(module22_87_opt)
+ opt_add_2176 = opt_split_2175 + self.add_2176_bias
+ opt_add_2177 = P.Add()(opt_add_2176, opt_split_2175_1)
+ module20_86_opt = self.module20_86(opt_add_2177)
+ opt_reshape_2187 = self.reshape_2187(module20_86_opt, self.reshape_2187_shape)
+ opt_split_2188, opt_split_2188_1 = self.split_2188(opt_reshape_2187)
+ opt_mul_2189 = P.Mul()(opt_split_2188, opt_split_2175)
+ opt_mul_2190 = P.Mul()(opt_split_2188_1, opt_split_2175_1)
+ opt_add_2191 = opt_mul_2189 + self.add_2191_bias
+ opt_add_2192 = P.Add()(opt_add_2191, opt_mul_2190)
+ opt_conv2d_2193 = self.conv2d_2193(opt_add_2192)
+ opt_add_2194 = P.Add()(opt_conv2d_2193, opt_relu_2170)
+ opt_relu_2195 = self.relu_2195(opt_add_2194)
+ module22_88_opt = self.module22_88(opt_relu_2195)
+ opt_split_2200, opt_split_2200_1 = self.split_2200(module22_88_opt)
+ opt_add_2201 = opt_split_2200 + self.add_2201_bias
+ opt_add_2202 = P.Add()(opt_add_2201, opt_split_2200_1)
+ module20_87_opt = self.module20_87(opt_add_2202)
+ opt_reshape_2212 = self.reshape_2212(module20_87_opt, self.reshape_2212_shape)
+ opt_split_2213, opt_split_2213_1 = self.split_2213(opt_reshape_2212)
+ opt_mul_2214 = P.Mul()(opt_split_2213, opt_split_2200)
+ opt_mul_2215 = P.Mul()(opt_split_2213_1, opt_split_2200_1)
+ opt_add_2216 = opt_mul_2214 + self.add_2216_bias
+ opt_add_2217 = P.Add()(opt_add_2216, opt_mul_2215)
+ opt_conv2d_2218 = self.conv2d_2218(opt_add_2217)
+ opt_add_2219 = P.Add()(opt_conv2d_2218, opt_relu_2195)
+ opt_relu_2220 = self.relu_2220(opt_add_2219)
+ module22_89_opt = self.module22_89(opt_relu_2220)
+ opt_split_2225, opt_split_2225_1 = self.split_2225(module22_89_opt)
+ opt_add_2226 = opt_split_2225 + self.add_2226_bias
+ opt_add_2227 = P.Add()(opt_add_2226, opt_split_2225_1)
+ module20_88_opt = self.module20_88(opt_add_2227)
+ opt_reshape_2237 = self.reshape_2237(module20_88_opt, self.reshape_2237_shape)
+ opt_split_2238, opt_split_2238_1 = self.split_2238(opt_reshape_2237)
+ opt_mul_2239 = P.Mul()(opt_split_2238, opt_split_2225)
+ opt_mul_2240 = P.Mul()(opt_split_2238_1, opt_split_2225_1)
+ opt_add_2241 = opt_mul_2239 + self.add_2241_bias
+ opt_add_2242 = P.Add()(opt_add_2241, opt_mul_2240)
+ opt_conv2d_2243 = self.conv2d_2243(opt_add_2242)
+ opt_add_2244 = P.Add()(opt_conv2d_2243, opt_relu_2220)
+ opt_relu_2245 = self.relu_2245(opt_add_2244)
+ opt_avgpool2d_2246 = self.avgpool2d_2246(opt_relu_2245)
+ opt_reshape_2247 = self.reshape_2247(opt_avgpool2d_2246, self.reshape_2247_shape)
+ opt_flatten_2248 = self.flatten_2248(opt_reshape_2247)
+ opt_dense_2249 = self.dense_2249(opt_flatten_2248)
+ return opt_dense_2249
diff --git a/research/cvtmodel/resnest/src/resnest50.py b/research/cvtmodel/resnest/src/resnest50.py
new file mode 100644
index 0000000000000000000000000000000000000000..5e09a53bba4793b9e3e2182c01ea9582aeee738c
--- /dev/null
+++ b/research/cvtmodel/resnest/src/resnest50.py
@@ -0,0 +1,1014 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module5(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_0_kernel_size, conv2d_0_padding,
+ conv2d_0_pad_mode, conv2d_0_group):
+ super(Module5, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=conv2d_0_kernel_size,
+ stride=(1, 1),
+ padding=conv2d_0_padding,
+ pad_mode=conv2d_0_pad_mode,
+ dilation=(1, 1),
+ group=conv2d_0_group,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ return opt_relu_1
+
+
+class Module22(nn.Cell):
+ def __init__(self, module5_0_conv2d_0_in_channels, module5_0_conv2d_0_out_channels, module5_0_conv2d_0_kernel_size,
+ module5_0_conv2d_0_padding, module5_0_conv2d_0_pad_mode, module5_0_conv2d_0_group,
+ module5_1_conv2d_0_in_channels, module5_1_conv2d_0_out_channels, module5_1_conv2d_0_kernel_size,
+ module5_1_conv2d_0_padding, module5_1_conv2d_0_pad_mode, module5_1_conv2d_0_group):
+ super(Module22, self).__init__()
+ self.module5_0 = Module5(conv2d_0_in_channels=module5_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_0_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_0_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_0_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_0_conv2d_0_pad_mode,
+ conv2d_0_group=module5_0_conv2d_0_group)
+ self.module5_1 = Module5(conv2d_0_in_channels=module5_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module5_1_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module5_1_conv2d_0_kernel_size,
+ conv2d_0_padding=module5_1_conv2d_0_padding,
+ conv2d_0_pad_mode=module5_1_conv2d_0_pad_mode,
+ conv2d_0_group=module5_1_conv2d_0_group)
+
+ def construct(self, x):
+ module5_0_opt = self.module5_0(x)
+ module5_1_opt = self.module5_1(module5_0_opt)
+ return module5_1_opt
+
+
+class Module0(nn.Cell):
+ def __init__(self, avgpool2d_0_kernel_size, conv2d_1_in_channels, conv2d_1_out_channels, conv2d_3_in_channels,
+ conv2d_3_out_channels, reshape_4_shape):
+ super(Module0, self).__init__()
+ self.avgpool2d_0 = nn.AvgPool2d(kernel_size=avgpool2d_0_kernel_size)
+ self.conv2d_1 = nn.Conv2d(in_channels=conv2d_1_in_channels,
+ out_channels=conv2d_1_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_2 = nn.ReLU()
+ self.conv2d_3 = nn.Conv2d(in_channels=conv2d_3_in_channels,
+ out_channels=conv2d_3_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.reshape_4 = P.Reshape()
+ self.reshape_4_shape = tuple(reshape_4_shape)
+ self.transpose_5 = P.Transpose()
+
+ def construct(self, x):
+ opt_avgpool2d_0 = self.avgpool2d_0(x)
+ opt_conv2d_1 = self.conv2d_1(opt_avgpool2d_0)
+ opt_relu_2 = self.relu_2(opt_conv2d_1)
+ opt_conv2d_3 = self.conv2d_3(opt_relu_2)
+ opt_reshape_4 = self.reshape_4(opt_conv2d_3, self.reshape_4_shape)
+ opt_transpose_5 = self.transpose_5(opt_reshape_4, (0, 3, 1, 2))
+ return opt_transpose_5
+
+
+class Module20(nn.Cell):
+ def __init__(self, reshape_2_shape, module0_0_avgpool2d_0_kernel_size, module0_0_conv2d_1_in_channels,
+ module0_0_conv2d_1_out_channels, module0_0_conv2d_3_in_channels, module0_0_conv2d_3_out_channels,
+ module0_0_reshape_4_shape):
+ super(Module20, self).__init__()
+ self.module0_0 = Module0(avgpool2d_0_kernel_size=module0_0_avgpool2d_0_kernel_size,
+ conv2d_1_in_channels=module0_0_conv2d_1_in_channels,
+ conv2d_1_out_channels=module0_0_conv2d_1_out_channels,
+ conv2d_3_in_channels=module0_0_conv2d_3_in_channels,
+ conv2d_3_out_channels=module0_0_conv2d_3_out_channels,
+ reshape_4_shape=module0_0_reshape_4_shape)
+ self.softmax_0 = nn.Softmax(axis=3)
+ self.transpose_1 = P.Transpose()
+ self.reshape_2 = P.Reshape()
+ self.reshape_2_shape = tuple(reshape_2_shape)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ opt_softmax_0 = self.softmax_0(module0_0_opt)
+ opt_transpose_1 = self.transpose_1(opt_softmax_0, (0, 3, 2, 1))
+ opt_reshape_2 = self.reshape_2(opt_transpose_1, self.reshape_2_shape)
+ return opt_reshape_2
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=32,
+ kernel_size=(3, 3),
+ stride=(2, 2),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.module22_0 = Module22(module5_0_conv2d_0_in_channels=32,
+ module5_0_conv2d_0_out_channels=32,
+ module5_0_conv2d_0_kernel_size=(3, 3),
+ module5_0_conv2d_0_padding=(1, 1, 1, 1),
+ module5_0_conv2d_0_pad_mode="pad",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=32,
+ module5_1_conv2d_0_out_channels=64,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=1)
+ self.pad_maxpool2d_6 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_6 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module22_1 = Module22(module5_0_conv2d_0_in_channels=64,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_13 = P.Split(axis=1, output_num=2)
+ self.add_14_bias = 0.0
+ self.module20_0 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(56, 56),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_25 = P.Reshape()
+ self.reshape_25_shape = tuple([1, 128, 1, 1])
+ self.split_26 = P.Split(axis=1, output_num=2)
+ self.add_29_bias = 0.0
+ self.conv2d_31 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_8 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_8 = nn.AvgPool2d(kernel_size=(1, 1), stride=(1, 1))
+ self.conv2d_10 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_33 = nn.ReLU()
+ self.module22_2 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_38 = P.Split(axis=1, output_num=2)
+ self.add_39_bias = 0.0
+ self.module20_1 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(56, 56),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_50 = P.Reshape()
+ self.reshape_50_shape = tuple([1, 128, 1, 1])
+ self.split_51 = P.Split(axis=1, output_num=2)
+ self.add_54_bias = 0.0
+ self.conv2d_56 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_58 = nn.ReLU()
+ self.module22_3 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=64,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=64,
+ module5_1_conv2d_0_out_channels=128,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_63 = P.Split(axis=1, output_num=2)
+ self.add_64_bias = 0.0
+ self.module20_2 = Module20(reshape_2_shape=[1, 128],
+ module0_0_avgpool2d_0_kernel_size=(56, 56),
+ module0_0_conv2d_1_in_channels=64,
+ module0_0_conv2d_1_out_channels=32,
+ module0_0_conv2d_3_in_channels=32,
+ module0_0_conv2d_3_out_channels=128,
+ module0_0_reshape_4_shape=[1, 1, 2, 64])
+ self.reshape_75 = P.Reshape()
+ self.reshape_75_shape = tuple([1, 128, 1, 1])
+ self.split_76 = P.Split(axis=1, output_num=2)
+ self.add_79_bias = 0.0
+ self.conv2d_81 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_83 = nn.ReLU()
+ self.module22_4 = Module22(module5_0_conv2d_0_in_channels=256,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_90 = P.Split(axis=1, output_num=2)
+ self.add_91_bias = 0.0
+ self.module20_3 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(56, 56),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_102 = P.Reshape()
+ self.reshape_102_shape = tuple([1, 256, 1, 1])
+ self.split_103 = P.Split(axis=1, output_num=2)
+ self.add_106_bias = 0.0
+ self.pad_108 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_109 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_109 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_110 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_85 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_85 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_87 = nn.Conv2d(in_channels=256,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_112 = nn.ReLU()
+ self.module22_5 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_117 = P.Split(axis=1, output_num=2)
+ self.add_118_bias = 0.0
+ self.module20_4 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(28, 28),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_129 = P.Reshape()
+ self.reshape_129_shape = tuple([1, 256, 1, 1])
+ self.split_130 = P.Split(axis=1, output_num=2)
+ self.add_133_bias = 0.0
+ self.conv2d_135 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_137 = nn.ReLU()
+ self.module22_6 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_142 = P.Split(axis=1, output_num=2)
+ self.add_143_bias = 0.0
+ self.module20_5 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(28, 28),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_154 = P.Reshape()
+ self.reshape_154_shape = tuple([1, 256, 1, 1])
+ self.split_155 = P.Split(axis=1, output_num=2)
+ self.add_158_bias = 0.0
+ self.conv2d_160 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_162 = nn.ReLU()
+ self.module22_7 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=128,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=128,
+ module5_1_conv2d_0_out_channels=256,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_167 = P.Split(axis=1, output_num=2)
+ self.add_168_bias = 0.0
+ self.module20_6 = Module20(reshape_2_shape=[1, 256],
+ module0_0_avgpool2d_0_kernel_size=(28, 28),
+ module0_0_conv2d_1_in_channels=128,
+ module0_0_conv2d_1_out_channels=64,
+ module0_0_conv2d_3_in_channels=64,
+ module0_0_conv2d_3_out_channels=256,
+ module0_0_reshape_4_shape=[1, 1, 2, 128])
+ self.reshape_179 = P.Reshape()
+ self.reshape_179_shape = tuple([1, 256, 1, 1])
+ self.split_180 = P.Split(axis=1, output_num=2)
+ self.add_183_bias = 0.0
+ self.conv2d_185 = nn.Conv2d(in_channels=128,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_187 = nn.ReLU()
+ self.module22_8 = Module22(module5_0_conv2d_0_in_channels=512,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_194 = P.Split(axis=1, output_num=2)
+ self.add_195_bias = 0.0
+ self.module20_7 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(28, 28),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_206 = P.Reshape()
+ self.reshape_206_shape = tuple([1, 512, 1, 1])
+ self.split_207 = P.Split(axis=1, output_num=2)
+ self.add_210_bias = 0.0
+ self.pad_212 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_213 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_213 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_214 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_189 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_189 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_191 = nn.Conv2d(in_channels=512,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_216 = nn.ReLU()
+ self.module22_9 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_221 = P.Split(axis=1, output_num=2)
+ self.add_222_bias = 0.0
+ self.module20_8 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_233 = P.Reshape()
+ self.reshape_233_shape = tuple([1, 512, 1, 1])
+ self.split_234 = P.Split(axis=1, output_num=2)
+ self.add_237_bias = 0.0
+ self.conv2d_239 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_241 = nn.ReLU()
+ self.module22_10 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_246 = P.Split(axis=1, output_num=2)
+ self.add_247_bias = 0.0
+ self.module20_9 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_258 = P.Reshape()
+ self.reshape_258_shape = tuple([1, 512, 1, 1])
+ self.split_259 = P.Split(axis=1, output_num=2)
+ self.add_262_bias = 0.0
+ self.conv2d_264 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_266 = nn.ReLU()
+ self.module22_11 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_271 = P.Split(axis=1, output_num=2)
+ self.add_272_bias = 0.0
+ self.module20_10 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_283 = P.Reshape()
+ self.reshape_283_shape = tuple([1, 512, 1, 1])
+ self.split_284 = P.Split(axis=1, output_num=2)
+ self.add_287_bias = 0.0
+ self.conv2d_289 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_291 = nn.ReLU()
+ self.module22_12 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_296 = P.Split(axis=1, output_num=2)
+ self.add_297_bias = 0.0
+ self.module20_11 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_308 = P.Reshape()
+ self.reshape_308_shape = tuple([1, 512, 1, 1])
+ self.split_309 = P.Split(axis=1, output_num=2)
+ self.add_312_bias = 0.0
+ self.conv2d_314 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_316 = nn.ReLU()
+ self.module22_13 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=256,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=256,
+ module5_1_conv2d_0_out_channels=512,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_321 = P.Split(axis=1, output_num=2)
+ self.add_322_bias = 0.0
+ self.module20_12 = Module20(reshape_2_shape=[1, 512],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=256,
+ module0_0_conv2d_1_out_channels=128,
+ module0_0_conv2d_3_in_channels=128,
+ module0_0_conv2d_3_out_channels=512,
+ module0_0_reshape_4_shape=[1, 1, 2, 256])
+ self.reshape_333 = P.Reshape()
+ self.reshape_333_shape = tuple([1, 512, 1, 1])
+ self.split_334 = P.Split(axis=1, output_num=2)
+ self.add_337_bias = 0.0
+ self.conv2d_339 = nn.Conv2d(in_channels=256,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_341 = nn.ReLU()
+ self.module22_14 = Module22(module5_0_conv2d_0_in_channels=1024,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_348 = P.Split(axis=1, output_num=2)
+ self.add_349_bias = 0.0
+ self.module20_13 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(14, 14),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_360 = P.Reshape()
+ self.reshape_360_shape = tuple([1, 1024, 1, 1])
+ self.split_361 = P.Split(axis=1, output_num=2)
+ self.add_364_bias = 0.0
+ self.pad_366 = nn.Pad(paddings=((0, 0), (0, 0), (1, 1), (1, 1)), mode="CONSTANT")
+ self.pad_avgpool2d_367 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_367 = nn.AvgPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.conv2d_368 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.pad_avgpool2d_343 = nn.Pad(paddings=((0, 0), (0, 0), (0, 0), (0, 0)))
+ self.avgpool2d_343 = nn.AvgPool2d(kernel_size=(2, 2), stride=(2, 2))
+ self.conv2d_345 = nn.Conv2d(in_channels=1024,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_370 = nn.ReLU()
+ self.module22_15 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_375 = P.Split(axis=1, output_num=2)
+ self.add_376_bias = 0.0
+ self.module20_14 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(7, 7),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_387 = P.Reshape()
+ self.reshape_387_shape = tuple([1, 1024, 1, 1])
+ self.split_388 = P.Split(axis=1, output_num=2)
+ self.add_391_bias = 0.0
+ self.conv2d_393 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_395 = nn.ReLU()
+ self.module22_16 = Module22(module5_0_conv2d_0_in_channels=2048,
+ module5_0_conv2d_0_out_channels=512,
+ module5_0_conv2d_0_kernel_size=(1, 1),
+ module5_0_conv2d_0_padding=0,
+ module5_0_conv2d_0_pad_mode="valid",
+ module5_0_conv2d_0_group=1,
+ module5_1_conv2d_0_in_channels=512,
+ module5_1_conv2d_0_out_channels=1024,
+ module5_1_conv2d_0_kernel_size=(3, 3),
+ module5_1_conv2d_0_padding=(1, 1, 1, 1),
+ module5_1_conv2d_0_pad_mode="pad",
+ module5_1_conv2d_0_group=2)
+ self.split_400 = P.Split(axis=1, output_num=2)
+ self.add_401_bias = 0.0
+ self.module20_15 = Module20(reshape_2_shape=[1, 1024],
+ module0_0_avgpool2d_0_kernel_size=(7, 7),
+ module0_0_conv2d_1_in_channels=512,
+ module0_0_conv2d_1_out_channels=256,
+ module0_0_conv2d_3_in_channels=256,
+ module0_0_conv2d_3_out_channels=1024,
+ module0_0_reshape_4_shape=[1, 1, 2, 512])
+ self.reshape_412 = P.Reshape()
+ self.reshape_412_shape = tuple([1, 1024, 1, 1])
+ self.split_413 = P.Split(axis=1, output_num=2)
+ self.add_416_bias = 0.0
+ self.conv2d_418 = nn.Conv2d(in_channels=512,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_420 = nn.ReLU()
+ self.avgpool2d_421 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.reshape_422 = P.Reshape()
+ self.reshape_422_shape = tuple([1, 2048])
+ self.flatten_423 = nn.Flatten()
+ self.dense_424 = nn.Dense(in_channels=2048, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ module22_0_opt = self.module22_0(opt_relu_1)
+ opt_maxpool2d_6 = self.pad_maxpool2d_6(module22_0_opt)
+ opt_maxpool2d_6 = self.maxpool2d_6(opt_maxpool2d_6)
+ module22_1_opt = self.module22_1(opt_maxpool2d_6)
+ opt_split_13, opt_split_13_1 = self.split_13(module22_1_opt)
+ opt_add_14 = opt_split_13 + self.add_14_bias
+ opt_add_15 = P.Add()(opt_add_14, opt_split_13_1)
+ module20_0_opt = self.module20_0(opt_add_15)
+ opt_reshape_25 = self.reshape_25(module20_0_opt, self.reshape_25_shape)
+ opt_split_26, opt_split_26_1 = self.split_26(opt_reshape_25)
+ opt_mul_27 = P.Mul()(opt_split_26, opt_split_13)
+ opt_mul_28 = P.Mul()(opt_split_26_1, opt_split_13_1)
+ opt_add_29 = opt_mul_27 + self.add_29_bias
+ opt_add_30 = P.Add()(opt_add_29, opt_mul_28)
+ opt_conv2d_31 = self.conv2d_31(opt_add_30)
+ opt_avgpool2d_8 = self.pad_avgpool2d_8(opt_maxpool2d_6)
+ opt_avgpool2d_8 = self.avgpool2d_8(opt_avgpool2d_8)
+ opt_conv2d_10 = self.conv2d_10(opt_avgpool2d_8)
+ opt_add_32 = P.Add()(opt_conv2d_31, opt_conv2d_10)
+ opt_relu_33 = self.relu_33(opt_add_32)
+ module22_2_opt = self.module22_2(opt_relu_33)
+ opt_split_38, opt_split_38_1 = self.split_38(module22_2_opt)
+ opt_add_39 = opt_split_38 + self.add_39_bias
+ opt_add_40 = P.Add()(opt_add_39, opt_split_38_1)
+ module20_1_opt = self.module20_1(opt_add_40)
+ opt_reshape_50 = self.reshape_50(module20_1_opt, self.reshape_50_shape)
+ opt_split_51, opt_split_51_1 = self.split_51(opt_reshape_50)
+ opt_mul_52 = P.Mul()(opt_split_51, opt_split_38)
+ opt_mul_53 = P.Mul()(opt_split_51_1, opt_split_38_1)
+ opt_add_54 = opt_mul_52 + self.add_54_bias
+ opt_add_55 = P.Add()(opt_add_54, opt_mul_53)
+ opt_conv2d_56 = self.conv2d_56(opt_add_55)
+ opt_add_57 = P.Add()(opt_conv2d_56, opt_relu_33)
+ opt_relu_58 = self.relu_58(opt_add_57)
+ module22_3_opt = self.module22_3(opt_relu_58)
+ opt_split_63, opt_split_63_1 = self.split_63(module22_3_opt)
+ opt_add_64 = opt_split_63 + self.add_64_bias
+ opt_add_65 = P.Add()(opt_add_64, opt_split_63_1)
+ module20_2_opt = self.module20_2(opt_add_65)
+ opt_reshape_75 = self.reshape_75(module20_2_opt, self.reshape_75_shape)
+ opt_split_76, opt_split_76_1 = self.split_76(opt_reshape_75)
+ opt_mul_77 = P.Mul()(opt_split_76, opt_split_63)
+ opt_mul_78 = P.Mul()(opt_split_76_1, opt_split_63_1)
+ opt_add_79 = opt_mul_77 + self.add_79_bias
+ opt_add_80 = P.Add()(opt_add_79, opt_mul_78)
+ opt_conv2d_81 = self.conv2d_81(opt_add_80)
+ opt_add_82 = P.Add()(opt_conv2d_81, opt_relu_58)
+ opt_relu_83 = self.relu_83(opt_add_82)
+ module22_4_opt = self.module22_4(opt_relu_83)
+ opt_split_90, opt_split_90_1 = self.split_90(module22_4_opt)
+ opt_add_91 = opt_split_90 + self.add_91_bias
+ opt_add_92 = P.Add()(opt_add_91, opt_split_90_1)
+ module20_3_opt = self.module20_3(opt_add_92)
+ opt_reshape_102 = self.reshape_102(module20_3_opt, self.reshape_102_shape)
+ opt_split_103, opt_split_103_1 = self.split_103(opt_reshape_102)
+ opt_mul_104 = P.Mul()(opt_split_103, opt_split_90)
+ opt_mul_105 = P.Mul()(opt_split_103_1, opt_split_90_1)
+ opt_add_106 = opt_mul_104 + self.add_106_bias
+ opt_add_107 = P.Add()(opt_add_106, opt_mul_105)
+ opt_pad_108 = self.pad_108(opt_add_107)
+ opt_avgpool2d_109 = self.pad_avgpool2d_109(opt_pad_108)
+ opt_avgpool2d_109 = self.avgpool2d_109(opt_avgpool2d_109)
+ opt_conv2d_110 = self.conv2d_110(opt_avgpool2d_109)
+ opt_avgpool2d_85 = self.pad_avgpool2d_85(opt_relu_83)
+ opt_avgpool2d_85 = self.avgpool2d_85(opt_avgpool2d_85)
+ opt_conv2d_87 = self.conv2d_87(opt_avgpool2d_85)
+ opt_add_111 = P.Add()(opt_conv2d_110, opt_conv2d_87)
+ opt_relu_112 = self.relu_112(opt_add_111)
+ module22_5_opt = self.module22_5(opt_relu_112)
+ opt_split_117, opt_split_117_1 = self.split_117(module22_5_opt)
+ opt_add_118 = opt_split_117 + self.add_118_bias
+ opt_add_119 = P.Add()(opt_add_118, opt_split_117_1)
+ module20_4_opt = self.module20_4(opt_add_119)
+ opt_reshape_129 = self.reshape_129(module20_4_opt, self.reshape_129_shape)
+ opt_split_130, opt_split_130_1 = self.split_130(opt_reshape_129)
+ opt_mul_131 = P.Mul()(opt_split_130, opt_split_117)
+ opt_mul_132 = P.Mul()(opt_split_130_1, opt_split_117_1)
+ opt_add_133 = opt_mul_131 + self.add_133_bias
+ opt_add_134 = P.Add()(opt_add_133, opt_mul_132)
+ opt_conv2d_135 = self.conv2d_135(opt_add_134)
+ opt_add_136 = P.Add()(opt_conv2d_135, opt_relu_112)
+ opt_relu_137 = self.relu_137(opt_add_136)
+ module22_6_opt = self.module22_6(opt_relu_137)
+ opt_split_142, opt_split_142_1 = self.split_142(module22_6_opt)
+ opt_add_143 = opt_split_142 + self.add_143_bias
+ opt_add_144 = P.Add()(opt_add_143, opt_split_142_1)
+ module20_5_opt = self.module20_5(opt_add_144)
+ opt_reshape_154 = self.reshape_154(module20_5_opt, self.reshape_154_shape)
+ opt_split_155, opt_split_155_1 = self.split_155(opt_reshape_154)
+ opt_mul_156 = P.Mul()(opt_split_155, opt_split_142)
+ opt_mul_157 = P.Mul()(opt_split_155_1, opt_split_142_1)
+ opt_add_158 = opt_mul_156 + self.add_158_bias
+ opt_add_159 = P.Add()(opt_add_158, opt_mul_157)
+ opt_conv2d_160 = self.conv2d_160(opt_add_159)
+ opt_add_161 = P.Add()(opt_conv2d_160, opt_relu_137)
+ opt_relu_162 = self.relu_162(opt_add_161)
+ module22_7_opt = self.module22_7(opt_relu_162)
+ opt_split_167, opt_split_167_1 = self.split_167(module22_7_opt)
+ opt_add_168 = opt_split_167 + self.add_168_bias
+ opt_add_169 = P.Add()(opt_add_168, opt_split_167_1)
+ module20_6_opt = self.module20_6(opt_add_169)
+ opt_reshape_179 = self.reshape_179(module20_6_opt, self.reshape_179_shape)
+ opt_split_180, opt_split_180_1 = self.split_180(opt_reshape_179)
+ opt_mul_181 = P.Mul()(opt_split_180, opt_split_167)
+ opt_mul_182 = P.Mul()(opt_split_180_1, opt_split_167_1)
+ opt_add_183 = opt_mul_181 + self.add_183_bias
+ opt_add_184 = P.Add()(opt_add_183, opt_mul_182)
+ opt_conv2d_185 = self.conv2d_185(opt_add_184)
+ opt_add_186 = P.Add()(opt_conv2d_185, opt_relu_162)
+ opt_relu_187 = self.relu_187(opt_add_186)
+ module22_8_opt = self.module22_8(opt_relu_187)
+ opt_split_194, opt_split_194_1 = self.split_194(module22_8_opt)
+ opt_add_195 = opt_split_194 + self.add_195_bias
+ opt_add_196 = P.Add()(opt_add_195, opt_split_194_1)
+ module20_7_opt = self.module20_7(opt_add_196)
+ opt_reshape_206 = self.reshape_206(module20_7_opt, self.reshape_206_shape)
+ opt_split_207, opt_split_207_1 = self.split_207(opt_reshape_206)
+ opt_mul_208 = P.Mul()(opt_split_207, opt_split_194)
+ opt_mul_209 = P.Mul()(opt_split_207_1, opt_split_194_1)
+ opt_add_210 = opt_mul_208 + self.add_210_bias
+ opt_add_211 = P.Add()(opt_add_210, opt_mul_209)
+ opt_pad_212 = self.pad_212(opt_add_211)
+ opt_avgpool2d_213 = self.pad_avgpool2d_213(opt_pad_212)
+ opt_avgpool2d_213 = self.avgpool2d_213(opt_avgpool2d_213)
+ opt_conv2d_214 = self.conv2d_214(opt_avgpool2d_213)
+ opt_avgpool2d_189 = self.pad_avgpool2d_189(opt_relu_187)
+ opt_avgpool2d_189 = self.avgpool2d_189(opt_avgpool2d_189)
+ opt_conv2d_191 = self.conv2d_191(opt_avgpool2d_189)
+ opt_add_215 = P.Add()(opt_conv2d_214, opt_conv2d_191)
+ opt_relu_216 = self.relu_216(opt_add_215)
+ module22_9_opt = self.module22_9(opt_relu_216)
+ opt_split_221, opt_split_221_1 = self.split_221(module22_9_opt)
+ opt_add_222 = opt_split_221 + self.add_222_bias
+ opt_add_223 = P.Add()(opt_add_222, opt_split_221_1)
+ module20_8_opt = self.module20_8(opt_add_223)
+ opt_reshape_233 = self.reshape_233(module20_8_opt, self.reshape_233_shape)
+ opt_split_234, opt_split_234_1 = self.split_234(opt_reshape_233)
+ opt_mul_235 = P.Mul()(opt_split_234, opt_split_221)
+ opt_mul_236 = P.Mul()(opt_split_234_1, opt_split_221_1)
+ opt_add_237 = opt_mul_235 + self.add_237_bias
+ opt_add_238 = P.Add()(opt_add_237, opt_mul_236)
+ opt_conv2d_239 = self.conv2d_239(opt_add_238)
+ opt_add_240 = P.Add()(opt_conv2d_239, opt_relu_216)
+ opt_relu_241 = self.relu_241(opt_add_240)
+ module22_10_opt = self.module22_10(opt_relu_241)
+ opt_split_246, opt_split_246_1 = self.split_246(module22_10_opt)
+ opt_add_247 = opt_split_246 + self.add_247_bias
+ opt_add_248 = P.Add()(opt_add_247, opt_split_246_1)
+ module20_9_opt = self.module20_9(opt_add_248)
+ opt_reshape_258 = self.reshape_258(module20_9_opt, self.reshape_258_shape)
+ opt_split_259, opt_split_259_1 = self.split_259(opt_reshape_258)
+ opt_mul_260 = P.Mul()(opt_split_259, opt_split_246)
+ opt_mul_261 = P.Mul()(opt_split_259_1, opt_split_246_1)
+ opt_add_262 = opt_mul_260 + self.add_262_bias
+ opt_add_263 = P.Add()(opt_add_262, opt_mul_261)
+ opt_conv2d_264 = self.conv2d_264(opt_add_263)
+ opt_add_265 = P.Add()(opt_conv2d_264, opt_relu_241)
+ opt_relu_266 = self.relu_266(opt_add_265)
+ module22_11_opt = self.module22_11(opt_relu_266)
+ opt_split_271, opt_split_271_1 = self.split_271(module22_11_opt)
+ opt_add_272 = opt_split_271 + self.add_272_bias
+ opt_add_273 = P.Add()(opt_add_272, opt_split_271_1)
+ module20_10_opt = self.module20_10(opt_add_273)
+ opt_reshape_283 = self.reshape_283(module20_10_opt, self.reshape_283_shape)
+ opt_split_284, opt_split_284_1 = self.split_284(opt_reshape_283)
+ opt_mul_285 = P.Mul()(opt_split_284, opt_split_271)
+ opt_mul_286 = P.Mul()(opt_split_284_1, opt_split_271_1)
+ opt_add_287 = opt_mul_285 + self.add_287_bias
+ opt_add_288 = P.Add()(opt_add_287, opt_mul_286)
+ opt_conv2d_289 = self.conv2d_289(opt_add_288)
+ opt_add_290 = P.Add()(opt_conv2d_289, opt_relu_266)
+ opt_relu_291 = self.relu_291(opt_add_290)
+ module22_12_opt = self.module22_12(opt_relu_291)
+ opt_split_296, opt_split_296_1 = self.split_296(module22_12_opt)
+ opt_add_297 = opt_split_296 + self.add_297_bias
+ opt_add_298 = P.Add()(opt_add_297, opt_split_296_1)
+ module20_11_opt = self.module20_11(opt_add_298)
+ opt_reshape_308 = self.reshape_308(module20_11_opt, self.reshape_308_shape)
+ opt_split_309, opt_split_309_1 = self.split_309(opt_reshape_308)
+ opt_mul_310 = P.Mul()(opt_split_309, opt_split_296)
+ opt_mul_311 = P.Mul()(opt_split_309_1, opt_split_296_1)
+ opt_add_312 = opt_mul_310 + self.add_312_bias
+ opt_add_313 = P.Add()(opt_add_312, opt_mul_311)
+ opt_conv2d_314 = self.conv2d_314(opt_add_313)
+ opt_add_315 = P.Add()(opt_conv2d_314, opt_relu_291)
+ opt_relu_316 = self.relu_316(opt_add_315)
+ module22_13_opt = self.module22_13(opt_relu_316)
+ opt_split_321, opt_split_321_1 = self.split_321(module22_13_opt)
+ opt_add_322 = opt_split_321 + self.add_322_bias
+ opt_add_323 = P.Add()(opt_add_322, opt_split_321_1)
+ module20_12_opt = self.module20_12(opt_add_323)
+ opt_reshape_333 = self.reshape_333(module20_12_opt, self.reshape_333_shape)
+ opt_split_334, opt_split_334_1 = self.split_334(opt_reshape_333)
+ opt_mul_335 = P.Mul()(opt_split_334, opt_split_321)
+ opt_mul_336 = P.Mul()(opt_split_334_1, opt_split_321_1)
+ opt_add_337 = opt_mul_335 + self.add_337_bias
+ opt_add_338 = P.Add()(opt_add_337, opt_mul_336)
+ opt_conv2d_339 = self.conv2d_339(opt_add_338)
+ opt_add_340 = P.Add()(opt_conv2d_339, opt_relu_316)
+ opt_relu_341 = self.relu_341(opt_add_340)
+ module22_14_opt = self.module22_14(opt_relu_341)
+ opt_split_348, opt_split_348_1 = self.split_348(module22_14_opt)
+ opt_add_349 = opt_split_348 + self.add_349_bias
+ opt_add_350 = P.Add()(opt_add_349, opt_split_348_1)
+ module20_13_opt = self.module20_13(opt_add_350)
+ opt_reshape_360 = self.reshape_360(module20_13_opt, self.reshape_360_shape)
+ opt_split_361, opt_split_361_1 = self.split_361(opt_reshape_360)
+ opt_mul_362 = P.Mul()(opt_split_361, opt_split_348)
+ opt_mul_363 = P.Mul()(opt_split_361_1, opt_split_348_1)
+ opt_add_364 = opt_mul_362 + self.add_364_bias
+ opt_add_365 = P.Add()(opt_add_364, opt_mul_363)
+ opt_pad_366 = self.pad_366(opt_add_365)
+ opt_avgpool2d_367 = self.pad_avgpool2d_367(opt_pad_366)
+ opt_avgpool2d_367 = self.avgpool2d_367(opt_avgpool2d_367)
+ opt_conv2d_368 = self.conv2d_368(opt_avgpool2d_367)
+ opt_avgpool2d_343 = self.pad_avgpool2d_343(opt_relu_341)
+ opt_avgpool2d_343 = self.avgpool2d_343(opt_avgpool2d_343)
+ opt_conv2d_345 = self.conv2d_345(opt_avgpool2d_343)
+ opt_add_369 = P.Add()(opt_conv2d_368, opt_conv2d_345)
+ opt_relu_370 = self.relu_370(opt_add_369)
+ module22_15_opt = self.module22_15(opt_relu_370)
+ opt_split_375, opt_split_375_1 = self.split_375(module22_15_opt)
+ opt_add_376 = opt_split_375 + self.add_376_bias
+ opt_add_377 = P.Add()(opt_add_376, opt_split_375_1)
+ module20_14_opt = self.module20_14(opt_add_377)
+ opt_reshape_387 = self.reshape_387(module20_14_opt, self.reshape_387_shape)
+ opt_split_388, opt_split_388_1 = self.split_388(opt_reshape_387)
+ opt_mul_389 = P.Mul()(opt_split_388, opt_split_375)
+ opt_mul_390 = P.Mul()(opt_split_388_1, opt_split_375_1)
+ opt_add_391 = opt_mul_389 + self.add_391_bias
+ opt_add_392 = P.Add()(opt_add_391, opt_mul_390)
+ opt_conv2d_393 = self.conv2d_393(opt_add_392)
+ opt_add_394 = P.Add()(opt_conv2d_393, opt_relu_370)
+ opt_relu_395 = self.relu_395(opt_add_394)
+ module22_16_opt = self.module22_16(opt_relu_395)
+ opt_split_400, opt_split_400_1 = self.split_400(module22_16_opt)
+ opt_add_401 = opt_split_400 + self.add_401_bias
+ opt_add_402 = P.Add()(opt_add_401, opt_split_400_1)
+ module20_15_opt = self.module20_15(opt_add_402)
+ opt_reshape_412 = self.reshape_412(module20_15_opt, self.reshape_412_shape)
+ opt_split_413, opt_split_413_1 = self.split_413(opt_reshape_412)
+ opt_mul_414 = P.Mul()(opt_split_413, opt_split_400)
+ opt_mul_415 = P.Mul()(opt_split_413_1, opt_split_400_1)
+ opt_add_416 = opt_mul_414 + self.add_416_bias
+ opt_add_417 = P.Add()(opt_add_416, opt_mul_415)
+ opt_conv2d_418 = self.conv2d_418(opt_add_417)
+ opt_add_419 = P.Add()(opt_conv2d_418, opt_relu_395)
+ opt_relu_420 = self.relu_420(opt_add_419)
+ opt_avgpool2d_421 = self.avgpool2d_421(opt_relu_420)
+ opt_reshape_422 = self.reshape_422(opt_avgpool2d_421, self.reshape_422_shape)
+ opt_flatten_423 = self.flatten_423(opt_reshape_422)
+ opt_dense_424 = self.dense_424(opt_flatten_423)
+ return opt_dense_424
diff --git a/research/cvtmodel/wide_resnet/README_CN.md b/research/cvtmodel/wide_resnet/README_CN.md
new file mode 100644
index 0000000000000000000000000000000000000000..06d4dbcc6f4247ca8c2222c1707576423e171fe2
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/README_CN.md
@@ -0,0 +1,101 @@
+# 目录
+
+<!-- TOC -->
+
+- [目录](#目录)
+- [Wide Resnet描述](#resnet描述)
+- [模型架构](#模型架构)
+- [数据集](#数据集)
+- [环境要求](#环境要求)
+- [脚本说明](#脚本说明)
+ - [脚本及样例代码](#脚本及样例代码)
+ - [导出过程](#导出过程)
+ - [导出](#导出)
+ - [推理过程](#推理过程)
+ - [推理](#推理)
+- [ModelZoo主页](#modelzoo主页)
+
+<!-- /TOC -->
+
+# Wide Resnet描述
+
+Wide Resnet是一个基于卷积的神经网络，用于图像分类。有关该模型的描述，可查阅(https://pytorch.org/hub/pytorch_vision_wide_resnet/)。
+本仓库中是基于torch提供的模型文件，使用MindConverter工具转化出来ckpt文件，进行全量推理以验证模型文件精度。
+
+# 模型架构
+
+Wide Resnet模型支持一种模式：Wide Resnet-101 V2。
+
+# 数据集
+
+Wide Resnet使用的数据集： ImageNet
+
+数据集的默认配置如下：
+
+- 测试数据集预处理：
+ - 图像的输入尺寸：224\*224（将图像缩放到256\*256，然后在中央区域裁剪图像）
+ - 根据平均值和标准偏差对输入图像进行归一化
+
+# 环境要求
+
+- 硬件（Ascend/GPU）
+- 准备Ascend或GPU处理器搭建硬件环境。
+- 框架
+- [MindSpore](https://www.mindspore.cn/install)
+- 如需查看详情，请参见如下资源：
+- [MindSpore教程](https://www.mindspore.cn/tutorials/zh-CN/master/index.html)
+- [MindSpore Python API](https://www.mindspore.cn/docs/api/zh-CN/master/index.html)
+
+# 脚本说明
+
+## 脚本及样例代码
+
+```shell
+├── model_zoo
+ ├── README.md // 所有模型的说明
+ ├── Wide Resnet
+ ├── README_CN.md // Wide Resnet相关说明
+ ├── ascend310_infer // 实现310推理源代码
+ ├── scripts
+ │ ├── run_infer_310.sh // Ascend 310 推理shell脚本
+ ├── src
+ │ ├── wide_resnet101_2.py // Wide Resnet-101 V2模型文件
+ ├── export.py // 导出脚本
+ ├── postprocess.py // 310 推理后处理脚本
+```
+
+## 导出过程
+
+### 导出
+
+```shell
+python export.py --backbone [NET_NAME] --ckpt_path [CKPT_PATH] --device_target [DEVICE_TARGET] --device_id 0 --file_format [EXPORT_FORMAT] --file_name [FILE_NAME]
+```
+
+`backbone` 可选 ["wideresnet101"]
+`EXPORT_FORMAT` 可选 ["AIR", "MINDIR"]
+
+## 推理过程
+
+### 推理
+
+在推理之前需要先导出模型，AIR模型只能在昇腾910环境上导出，MINDIR可以在任意环境上导出。
+
+```shell
+# 昇腾310 推理
+bash run_infer_310.sh [MINDIR_PATH] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+```
+
+-注: Wide-Resnet系列网络使用ImageNet数据集,图片的label是将文件夹排序后从0开始编号所得的数字.
+
+推理的结果保存在当前目录下，在acc.log日志文件中可以找到类似以下的结果。
+Wide Resnet101 V2网络使用ImageNet推理得到的结果如下:
+
+ ```log
+ after allreduce eval: top1_correct=38848, tot=50000, acc=77.70%
+ after allreduce eval: top5_correct=46861, tot=50000, acc=93.72%
+ ```
+
+# ModelZoo主页
+
+ 请浏览官网[主页](https://gitee.com/mindspore/models)。
diff --git a/research/cvtmodel/wide_resnet/ascend310_infer/CMakeLists.txt b/research/cvtmodel/wide_resnet/ascend310_infer/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..ee3c85447340e0449ff2b70ed24f60a17e07b2b6
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/ascend310_infer/CMakeLists.txt
@@ -0,0 +1,14 @@
+cmake_minimum_required(VERSION 3.14.1)
+project(Ascend310Infer)
+add_compile_definitions(_GLIBCXX_USE_CXX11_ABI=0)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O0 -g -std=c++17 -Werror -Wall -fPIE -Wl,--allow-shlib-undefined")
+set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
+option(MINDSPORE_PATH "mindspore install path" "")
+include_directories(${MINDSPORE_PATH})
+include_directories(${MINDSPORE_PATH}/include)
+include_directories(${PROJECT_SRC_ROOT})
+find_library(MS_LIB libmindspore.so ${MINDSPORE_PATH}/lib)
+file(GLOB_RECURSE MD_LIB ${MINDSPORE_PATH}/_c_dataengine*)
+
+add_executable(main src/main.cc src/utils.cc)
+target_link_libraries(main ${MS_LIB} ${MD_LIB} gflags)
diff --git a/research/cvtmodel/wide_resnet/ascend310_infer/build.sh b/research/cvtmodel/wide_resnet/ascend310_infer/build.sh
new file mode 100644
index 0000000000000000000000000000000000000000..770a8851efade7f352039fc8665d307ae1abbb00
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/ascend310_infer/build.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [ ! -d out ]; then
+ mkdir out
+fi
+cd out || exit
+cmake .. \
+ -DMINDSPORE_PATH="`pip show mindspore-ascend | grep Location | awk '{print $2"/mindspore"}' | xargs realpath`"
+make
diff --git a/research/cvtmodel/wide_resnet/ascend310_infer/inc/utils.h b/research/cvtmodel/wide_resnet/ascend310_infer/inc/utils.h
new file mode 100644
index 0000000000000000000000000000000000000000..efebe03a8c1179f5a1f9d5f7ee07e0352a9937c6
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/ascend310_infer/inc/utils.h
@@ -0,0 +1,32 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef MINDSPORE_INFERENCE_UTILS_H_
+#define MINDSPORE_INFERENCE_UTILS_H_
+
+#include <sys/stat.h>
+#include <dirent.h>
+#include <vector>
+#include <string>
+#include <memory>
+#include "include/api/types.h"
+
+std::vector<std::string> GetAllFiles(std::string_view dirName);
+DIR *OpenDir(std::string_view dirName);
+std::string RealPath(std::string_view path);
+mindspore::MSTensor ReadFileToTensor(const std::string &file);
+int WriteResult(const std::string& imageFile, const std::vector<mindspore::MSTensor> &outputs);
+#endif
diff --git a/research/cvtmodel/wide_resnet/ascend310_infer/src/main.cc b/research/cvtmodel/wide_resnet/ascend310_infer/src/main.cc
new file mode 100644
index 0000000000000000000000000000000000000000..a3885312e2690acf34137bfd4490dd0401c21de4
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/ascend310_infer/src/main.cc
@@ -0,0 +1,166 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <sys/time.h>
+#include <gflags/gflags.h>
+#include <dirent.h>
+#include <iostream>
+#include <string>
+#include <algorithm>
+#include <iosfwd>
+#include <vector>
+#include <fstream>
+
+#include "../inc/utils.h"
+#include "include/dataset/execute.h"
+#include "include/dataset/transforms.h"
+#include "include/dataset/vision.h"
+#include "include/dataset/vision_ascend.h"
+#include "include/api/types.h"
+#include "include/api/model.h"
+#include "include/api/serialization.h"
+#include "include/api/context.h"
+
+using mindspore::Serialization;
+using mindspore::Model;
+using mindspore::Context;
+using mindspore::Status;
+using mindspore::ModelType;
+using mindspore::Graph;
+using mindspore::GraphCell;
+using mindspore::kSuccess;
+using mindspore::MSTensor;
+using mindspore::DataType;
+using mindspore::dataset::Execute;
+using mindspore::dataset::TensorTransform;
+using mindspore::dataset::vision::Decode;
+using mindspore::dataset::vision::Resize;
+using mindspore::dataset::vision::CenterCrop;
+using mindspore::dataset::vision::Normalize;
+using mindspore::dataset::vision::HWC2CHW;
+
+using mindspore::dataset::transforms::TypeCast;
+
+
+DEFINE_string(model_path, "", "model path");
+DEFINE_string(dataset, "ImageNet", "dataset: ImageNet or Cifar10");
+DEFINE_string(dataset_path, ".", "dataset path");
+DEFINE_int32(device_id, 0, "device id");
+
+int main(int argc, char **argv) {
+ gflags::ParseCommandLineFlags(&argc, &argv, true);
+ if (RealPath(FLAGS_model_path).empty()) {
+ std::cout << "Invalid model" << std::endl;
+ return 1;
+ }
+
+ std::transform(FLAGS_dataset.begin(), FLAGS_dataset.end(), FLAGS_dataset.begin(), ::tolower);
+
+ auto context = std::make_shared<Context>();
+ auto ascend310_info = std::make_shared<mindspore::Ascend310DeviceInfo>();
+ ascend310_info->SetDeviceID(FLAGS_device_id);
+ context->MutableDeviceInfo().push_back(ascend310_info);
+
+ Graph graph;
+ Status ret = Serialization::Load(FLAGS_model_path, ModelType::kMindIR, &graph);
+ if (ret != kSuccess) {
+ std::cout << "Load model failed." << std::endl;
+ return 1;
+ }
+
+ Model model;
+ ret = model.Build(GraphCell(graph), context);
+ if (ret != kSuccess) {
+ std::cout << "ERROR: Build failed." << std::endl;
+ return 1;
+ }
+
+ std::vector<MSTensor> modelInputs = model.GetInputs();
+
+ auto all_files = GetAllFiles(FLAGS_dataset_path);
+ if (all_files.empty()) {
+ std::cout << "ERROR: no input data." << std::endl;
+ return 1;
+ }
+
+ std::shared_ptr<TensorTransform> decode(new Decode());
+ std::shared_ptr<TensorTransform> resize(new Resize({256, 256}));
+ std::shared_ptr<TensorTransform> centerCrop(new CenterCrop({224, 224}));
+ std::shared_ptr<TensorTransform> normImageNet(new Normalize({123.675, 116.28, 103.53}, {58.395, 57.12, 57.375}));
+ std::shared_ptr<TensorTransform> normCifar10(new Normalize({127.5, 127.5, 127.5}, {127.5, 127.5, 127.5}));
+ std::shared_ptr<TensorTransform> hwc2chw(new HWC2CHW());
+
+ mindspore::dataset::Execute transformImageNet({decode, resize, centerCrop, normImageNet, hwc2chw});
+ mindspore::dataset::Execute transformCifar10({decode, normCifar10, hwc2chw});
+
+ std::map<double, double> costTime_map;
+
+ size_t size = all_files.size();
+ for (size_t i = 0; i < size; ++i) {
+ struct timeval start;
+ struct timeval end;
+ double startTime_ms;
+ double endTime_ms;
+ std::vector<MSTensor> inputs;
+ std::vector<MSTensor> outputs;
+
+ std::cout << "Start predict input files:" << all_files[i] << std::endl;
+ mindspore::MSTensor image = ReadFileToTensor(all_files[i]);
+
+ if (FLAGS_dataset.compare("cifar10") == 0) {
+ transformCifar10(image, &image);
+ } else if (FLAGS_dataset.compare("imagenet") == 0) {
+ transformImageNet(image, &image);
+ } else {
+ std::cout << "unsupported dataset ...";
+ return 1;
+ }
+
+ inputs.emplace_back(modelInputs[0].Name(), modelInputs[0].DataType(), modelInputs[0].Shape(),
+ image.Data().get(), image.DataSize());
+
+ gettimeofday(&start, NULL);
+ model.Predict(inputs, &outputs);
+ gettimeofday(&end, NULL);
+
+ startTime_ms = (1.0 * start.tv_sec * 1000000 + start.tv_usec) / 1000;
+ endTime_ms = (1.0 * end.tv_sec * 1000000 + end.tv_usec) / 1000;
+ costTime_map.insert(std::pair<double, double>(startTime_ms, endTime_ms));
+ int rst = WriteResult(all_files[i], outputs);
+ if (rst != 0) {
+ std::cout << "write result failed." << std::endl;
+ return rst;
+ }
+ }
+ double average = 0.0;
+ int infer_cnt = 0;
+ char tmpCh[256] = {0};
+ for (auto iter = costTime_map.begin(); iter != costTime_map.end(); iter++) {
+ double diff = 0.0;
+ diff = iter->second - iter->first;
+ average += diff;
+ infer_cnt++;
+ }
+
+ average = average/infer_cnt;
+ std::cout << "NN inference cost average time: "<< average << "ms of infer_count " << infer_cnt << std::endl;
+ std::string file_name = "./time_Result" + std::string("/test_perform_static.txt");
+ std::ofstream file_stream(file_name.c_str(), std::ios::trunc);
+ file_stream << tmpCh;
+ file_stream.close();
+ costTime_map.clear();
+ return 0;
+}
diff --git a/research/cvtmodel/wide_resnet/ascend310_infer/src/utils.cc b/research/cvtmodel/wide_resnet/ascend310_infer/src/utils.cc
new file mode 100644
index 0000000000000000000000000000000000000000..653b1de44962614ac77d44481f8d1a7bde52caaf
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/ascend310_infer/src/utils.cc
@@ -0,0 +1,160 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "inc/utils.h"
+
+#include <fstream>
+#include <algorithm>
+#include <iostream>
+
+using mindspore::MSTensor;
+using mindspore::DataType;
+
+std::vector<std::string> GetAllFiles(std::string_view dirName) {
+ struct dirent *filename;
+ DIR *dir = OpenDir(dirName);
+ if (dir == nullptr) {
+ return {};
+ }
+ std::vector<std::string> dirs;
+ std::vector<std::string> files;
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == "..") {
+ continue;
+ } else if (filename->d_type == DT_DIR) {
+ dirs.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else if (filename->d_type == DT_REG) {
+ files.emplace_back(std::string(dirName) + "/" + filename->d_name);
+ } else {
+ continue;
+ }
+ }
+
+ for (auto d : dirs) {
+ dir = OpenDir(d);
+ while ((filename = readdir(dir)) != nullptr) {
+ std::string dName = std::string(filename->d_name);
+ if (dName == "." || dName == ".." || filename->d_type != DT_REG) {
+ continue;
+ }
+ files.emplace_back(std::string(d) + "/" + filename->d_name);
+ }
+ }
+ std::sort(files.begin(), files.end());
+ for (auto &f : files) {
+ std::cout << "image file: " << f << std::endl;
+ }
+ return files;
+}
+
+int WriteResult(const std::string& imageFile, const std::vector<MSTensor> &outputs) {
+ std::string homePath = "./result_Files";
+ const int INVALID_POINTER = -1;
+ const int ERROR = -2;
+ for (size_t i = 0; i < outputs.size(); ++i) {
+ size_t outputSize;
+ std::shared_ptr<const void> netOutput;
+ netOutput = outputs[i].Data();
+ outputSize = outputs[i].DataSize();
+ int pos = imageFile.rfind('/');
+ std::string fileName(imageFile, pos + 1);
+ fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), '_' + std::to_string(i) + ".bin");
+ std::string outFileName = homePath + "/" + fileName;
+ FILE * outputFile = fopen(outFileName.c_str(), "wb");
+ if (outputFile == nullptr) {
+ std::cout << "open result file " << outFileName << " failed" << std::endl;
+ return INVALID_POINTER;
+ }
+ size_t size = fwrite(netOutput.get(), sizeof(char), outputSize, outputFile);
+ if (size != outputSize) {
+ fclose(outputFile);
+ outputFile = nullptr;
+ std::cout << "write result file " << outFileName << " failed, write size[" << size <<
+ "] is smaller than output size[" << outputSize << "], maybe the disk is full." << std::endl;
+ return ERROR;
+ }
+ fclose(outputFile);
+ outputFile = nullptr;
+ }
+ return 0;
+}
+
+mindspore::MSTensor ReadFileToTensor(const std::string &file) {
+ if (file.empty()) {
+ std::cout << "Pointer file is nullptr" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ std::ifstream ifs(file);
+ if (!ifs.good()) {
+ std::cout << "File: " << file << " is not exist" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ if (!ifs.is_open()) {
+ std::cout << "File: " << file << "open failed" << std::endl;
+ return mindspore::MSTensor();
+ }
+
+ ifs.seekg(0, std::ios::end);
+ size_t size = ifs.tellg();
+ mindspore::MSTensor buffer(file, mindspore::DataType::kNumberTypeUInt8, {static_cast<int64_t>(size)}, nullptr, size);
+
+ ifs.seekg(0, std::ios::beg);
+ ifs.read(reinterpret_cast<char *>(buffer.MutableData()), size);
+ ifs.close();
+
+ return buffer;
+}
+
+
+DIR *OpenDir(std::string_view dirName) {
+ if (dirName.empty()) {
+ std::cout << " dirName is null ! " << std::endl;
+ return nullptr;
+ }
+ std::string realPath = RealPath(dirName);
+ struct stat s;
+ lstat(realPath.c_str(), &s);
+ if (!S_ISDIR(s.st_mode)) {
+ std::cout << "dirName is not a valid directory !" << std::endl;
+ return nullptr;
+ }
+ DIR *dir;
+ dir = opendir(realPath.c_str());
+ if (dir == nullptr) {
+ std::cout << "Can not open dir " << dirName << std::endl;
+ return nullptr;
+ }
+ std::cout << "Successfully opened the dir " << dirName << std::endl;
+ return dir;
+}
+
+std::string RealPath(std::string_view path) {
+ char realPathMem[PATH_MAX] = {0};
+ char *realPathRet = nullptr;
+ realPathRet = realpath(path.data(), realPathMem);
+
+ if (realPathRet == nullptr) {
+ std::cout << "File: " << path << " is not exist.";
+ return "";
+ }
+
+ std::string realPath(realPathMem);
+ std::cout << path << " realpath is: " << realPath << std::endl;
+ return realPath;
+}
diff --git a/research/cvtmodel/wide_resnet/export.py b/research/cvtmodel/wide_resnet/export.py
new file mode 100644
index 0000000000000000000000000000000000000000..6d1ce3d38d22261b933615731880c6086242d52c
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/export.py
@@ -0,0 +1,53 @@
+# Copyright 2020-2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+"""export checkpoint file into air, onnx, mindir models
+ Suggest run as python export.py --file_name [file_name] --ckpt_files [ckpt path] --file_format [file format]
+"""
+import argparse
+import numpy as np
+import mindspore as ms
+from mindspore import context, Tensor
+from mindspore.train.serialization import export, load_checkpoint, load_param_into_net
+
+parser = argparse.ArgumentParser(description='post process for 310 inference')
+parser.add_argument("--backbone", type=str, required=True, default="wideresnet101V2", help="model backbone")
+parser.add_argument("--ckpt_path", type=str, required=True, help="checkpoint file path")
+parser.add_argument("--file_name", type=str, default="wideresnet101V2", help="file name")
+parser.add_argument("--file_format", type=str, default="MINDIR", choices=["MINDIR", "AIR"], help="file format")
+parser.add_argument("--device_target", type=str, default="Ascend", choices=["Ascend", "GPU"], help="device target")
+parser.add_argument("--device_id", type=int, default=0, help="device target")
+args = parser.parse_args()
+
+context.set_context(mode=context.GRAPH_MODE, device_target=args.device_target)
+
+def model_export():
+ '''main export function'''
+ if args.device_target == "Ascend":
+ context.set_context(device_id=args.device_id)
+ if args.backbone == "wideresnet101":
+ from src.wide_resnet101_2 import MainModel
+ image_size = 224
+
+ net = MainModel()
+
+ param_dict = load_checkpoint(args.ckpt_path)
+ load_param_into_net(net, param_dict)
+
+ input_arr = Tensor(np.zeros([1, 3, image_size, image_size]), ms.float32)
+ export(net, input_arr, file_name=args.file_name, file_format=args.file_format)
+
+if __name__ == '__main__':
+ model_export()
diff --git a/research/cvtmodel/wide_resnet/postprocess.py b/research/cvtmodel/wide_resnet/postprocess.py
new file mode 100644
index 0000000000000000000000000000000000000000..0bfa5b4a79ab40025bdb1344cfe30cf625743812
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/postprocess.py
@@ -0,0 +1,82 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+'''post process for 310 inference'''
+import os
+import argparse
+import numpy as np
+
+parser = argparse.ArgumentParser(description='post process for resnest 310 inference')
+parser.add_argument("--dataset", type=str, default="imagenet", help="result file path")
+parser.add_argument("--result_path", type=str, required=True, help="result file path")
+parser.add_argument("--label_file", type=str, required=True, help="label file")
+args = parser.parse_args()
+
+def get_top5_acc(top5_arg, gt_class):
+ sub_count = 0
+ for top5, gt in zip(top5_arg, gt_class):
+ if gt in top5:
+ sub_count += 1
+ return sub_count
+
+def read_label(label_file):
+ '''read label file'''
+ f = open(label_file, "r")
+ lines = f.readlines()
+
+ img_label = {}
+ for line in lines:
+ img_id = line.split(":")[0]
+ label = line.split(":")[1]
+ img_label[img_id] = label
+
+ return img_label
+
+def cal_acc(dataset, result_path, label_file):
+ '''cal acc function'''
+ img_label = read_label(label_file)
+
+ img_tot = 0
+ top1_correct = 0
+ top5_correct = 0
+
+ files = os.listdir(result_path)
+ for file in files:
+ full_file_path = os.path.join(result_path, file)
+ if os.path.isfile(full_file_path):
+ result = np.fromfile(full_file_path, dtype=np.float32).reshape(1, 1000)
+ gt_classes = int(img_label[file[:-6]])
+
+ top1_output = np.argmax(result, (-1))
+ top5_output = np.argsort(result)[:, -5:]
+
+ t1_correct = np.equal(top1_output, gt_classes).sum()
+ top1_correct += t1_correct
+ top5_correct += get_top5_acc(top5_output, [gt_classes])
+ img_tot += 1
+
+ results = [[top1_correct], [top5_correct], [img_tot]]
+
+ results = np.array(results)
+ top1_correct = results[0, 0]
+ top5_correct = results[1, 0]
+ img_tot = results[2, 0]
+ acc1 = 100.0 * top1_correct / img_tot
+ acc5 = 100.0 * top5_correct / img_tot
+ print('after allreduce eval: top1_correct={}, tot={}, acc={:.2f}%'.format(top1_correct, img_tot, acc1))
+ if dataset == 'imagenet':
+ print('after allreduce eval: top5_correct={}, tot={}, acc={:.2f}%'.format(top5_correct, img_tot, acc5))
+
+if __name__ == "__main__":
+ cal_acc(args.dataset, args.result_path, args.label_file)
diff --git a/research/cvtmodel/wide_resnet/scripts/run_infer_310.sh b/research/cvtmodel/wide_resnet/scripts/run_infer_310.sh
new file mode 100644
index 0000000000000000000000000000000000000000..94595d3663f8f98fc7f0fd280d067c4996289bf5
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/scripts/run_infer_310.sh
@@ -0,0 +1,106 @@
+#!/bin/bash
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+if [[ $# -lt 4 || $# -gt 5 ]]; then
+ echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [DATASET] [DATA_PATH] [LABEL_FILE] [DEVICE_ID]
+ DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
+exit 1
+fi
+
+get_real_path(){
+ if [ "${1:0:1}" == "/" ]; then
+ echo "$1"
+ else
+ echo "$(realpath -m $PWD/$1)"
+ fi
+}
+
+model=$(get_real_path $1)
+dataset=$2
+data_path=$(get_real_path $3)
+label_file=$(get_real_path $4)
+
+device_id=0
+
+if [ $# == 5 ]; then
+ device_id=$5
+fi
+
+echo $model
+echo $dataset
+echo $data_path
+echo $label_file
+echo $device_id
+
+export ASCEND_HOME=/usr/local/Ascend/
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+ export PATH=$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export TBE_IMPL_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+ export PYTHONPATH=${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+ export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+ export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+ export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+ export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi
+
+function compile_app()
+{
+ cd ../ascend310_infer || exit
+ if [ -f "Makefile" ]; then
+ make clean
+ fi
+ sh build.sh &> build.log
+
+ if [ $? -ne 0 ]; then
+ echo "compile app code failed"
+ exit 1
+ fi
+ cd - || exit
+}
+
+function infer()
+{
+ if [ -d result_Files ]; then
+ rm -rf ./result_Files
+ fi
+ if [ -d time_Result ]; then
+ rm -rf ./time_Result
+ fi
+ mkdir result_Files
+ mkdir time_Result
+ ../ascend310_infer/out/main --model_path=$model --dataset=$dataset --dataset_path=$data_path --device_id=$device_id &> infer.log
+
+ if [ $? -ne 0 ]; then
+ echo "execute inference failed"
+ exit 1
+ fi
+}
+
+function cal_acc()
+{
+ python ../postprocess.py --dataset=$dataset --label_file=$label_file --result_path=result_Files &> acc.log
+ if [ $? -ne 0 ]; then
+ echo "calculate accuracy failed"
+ exit 1
+ fi
+}
+
+compile_app
+infer
+cal_acc
diff --git a/research/cvtmodel/wide_resnet/src/wide_resnet101_2.py b/research/cvtmodel/wide_resnet/src/wide_resnet101_2.py
new file mode 100644
index 0000000000000000000000000000000000000000..08cbde609b9f8c7d0e76596c2d48d0198369edda
--- /dev/null
+++ b/research/cvtmodel/wide_resnet/src/wide_resnet101_2.py
@@ -0,0 +1,576 @@
+import mindspore.ops as P
+from mindspore import nn
+
+
+class Module3(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_0_kernel_size, conv2d_0_stride,
+ conv2d_0_padding, conv2d_0_pad_mode):
+ super(Module3, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=conv2d_0_kernel_size,
+ stride=conv2d_0_stride,
+ padding=conv2d_0_padding,
+ pad_mode=conv2d_0_pad_mode,
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ return opt_relu_1
+
+
+class Module8(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, module3_0_conv2d_0_in_channels,
+ module3_0_conv2d_0_out_channels, module3_0_conv2d_0_kernel_size, module3_0_conv2d_0_stride,
+ module3_0_conv2d_0_padding, module3_0_conv2d_0_pad_mode, module3_1_conv2d_0_in_channels,
+ module3_1_conv2d_0_out_channels, module3_1_conv2d_0_kernel_size, module3_1_conv2d_0_stride,
+ module3_1_conv2d_0_padding, module3_1_conv2d_0_pad_mode):
+ super(Module8, self).__init__()
+ self.module3_0 = Module3(conv2d_0_in_channels=module3_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module3_0_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module3_0_conv2d_0_kernel_size,
+ conv2d_0_stride=module3_0_conv2d_0_stride,
+ conv2d_0_padding=module3_0_conv2d_0_padding,
+ conv2d_0_pad_mode=module3_0_conv2d_0_pad_mode)
+ self.module3_1 = Module3(conv2d_0_in_channels=module3_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module3_1_conv2d_0_out_channels,
+ conv2d_0_kernel_size=module3_1_conv2d_0_kernel_size,
+ conv2d_0_stride=module3_1_conv2d_0_stride,
+ conv2d_0_padding=module3_1_conv2d_0_padding,
+ conv2d_0_pad_mode=module3_1_conv2d_0_pad_mode)
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+
+ def construct(self, x):
+ module3_0_opt = self.module3_0(x)
+ module3_1_opt = self.module3_1(module3_0_opt)
+ opt_conv2d_0 = self.conv2d_0(module3_1_opt)
+ return opt_conv2d_0
+
+
+class Module0(nn.Cell):
+ def __init__(self, conv2d_0_in_channels, conv2d_0_out_channels, conv2d_2_in_channels, conv2d_2_out_channels,
+ conv2d_4_in_channels, conv2d_4_out_channels):
+ super(Module0, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=conv2d_0_in_channels,
+ out_channels=conv2d_0_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.conv2d_2 = nn.Conv2d(in_channels=conv2d_2_in_channels,
+ out_channels=conv2d_2_out_channels,
+ kernel_size=(3, 3),
+ stride=(1, 1),
+ padding=(1, 1, 1, 1),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_3 = nn.ReLU()
+ self.conv2d_4 = nn.Conv2d(in_channels=conv2d_4_in_channels,
+ out_channels=conv2d_4_out_channels,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_6 = nn.ReLU()
+
+ def construct(self, x):
+ opt_conv2d_0 = self.conv2d_0(x)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_conv2d_2 = self.conv2d_2(opt_relu_1)
+ opt_relu_3 = self.relu_3(opt_conv2d_2)
+ opt_conv2d_4 = self.conv2d_4(opt_relu_3)
+ opt_add_5 = P.Add()(opt_conv2d_4, x)
+ opt_relu_6 = self.relu_6(opt_add_5)
+ return opt_relu_6
+
+
+class Module9(nn.Cell):
+ def __init__(self, module0_0_conv2d_0_in_channels, module0_0_conv2d_0_out_channels, module0_0_conv2d_2_in_channels,
+ module0_0_conv2d_2_out_channels, module0_0_conv2d_4_in_channels, module0_0_conv2d_4_out_channels,
+ module0_1_conv2d_0_in_channels, module0_1_conv2d_0_out_channels, module0_1_conv2d_2_in_channels,
+ module0_1_conv2d_2_out_channels, module0_1_conv2d_4_in_channels, module0_1_conv2d_4_out_channels):
+ super(Module9, self).__init__()
+ self.module0_0 = Module0(conv2d_0_in_channels=module0_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_0_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_0_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_0_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_0_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_0_conv2d_4_out_channels)
+ self.module0_1 = Module0(conv2d_0_in_channels=module0_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_1_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_1_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_1_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_1_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_1_conv2d_4_out_channels)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ module0_1_opt = self.module0_1(module0_0_opt)
+ return module0_1_opt
+
+
+class Module13(nn.Cell):
+ def __init__(self, module0_0_conv2d_0_in_channels, module0_0_conv2d_0_out_channels, module0_0_conv2d_2_in_channels,
+ module0_0_conv2d_2_out_channels, module0_0_conv2d_4_in_channels, module0_0_conv2d_4_out_channels,
+ module0_1_conv2d_0_in_channels, module0_1_conv2d_0_out_channels, module0_1_conv2d_2_in_channels,
+ module0_1_conv2d_2_out_channels, module0_1_conv2d_4_in_channels, module0_1_conv2d_4_out_channels,
+ module0_2_conv2d_0_in_channels, module0_2_conv2d_0_out_channels, module0_2_conv2d_2_in_channels,
+ module0_2_conv2d_2_out_channels, module0_2_conv2d_4_in_channels, module0_2_conv2d_4_out_channels,
+ module0_3_conv2d_0_in_channels, module0_3_conv2d_0_out_channels, module0_3_conv2d_2_in_channels,
+ module0_3_conv2d_2_out_channels, module0_3_conv2d_4_in_channels, module0_3_conv2d_4_out_channels,
+ module0_4_conv2d_0_in_channels, module0_4_conv2d_0_out_channels, module0_4_conv2d_2_in_channels,
+ module0_4_conv2d_2_out_channels, module0_4_conv2d_4_in_channels, module0_4_conv2d_4_out_channels,
+ module0_5_conv2d_0_in_channels, module0_5_conv2d_0_out_channels, module0_5_conv2d_2_in_channels,
+ module0_5_conv2d_2_out_channels, module0_5_conv2d_4_in_channels, module0_5_conv2d_4_out_channels,
+ module0_6_conv2d_0_in_channels, module0_6_conv2d_0_out_channels, module0_6_conv2d_2_in_channels,
+ module0_6_conv2d_2_out_channels, module0_6_conv2d_4_in_channels, module0_6_conv2d_4_out_channels,
+ module0_7_conv2d_0_in_channels, module0_7_conv2d_0_out_channels, module0_7_conv2d_2_in_channels,
+ module0_7_conv2d_2_out_channels, module0_7_conv2d_4_in_channels, module0_7_conv2d_4_out_channels):
+ super(Module13, self).__init__()
+ self.module0_0 = Module0(conv2d_0_in_channels=module0_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_0_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_0_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_0_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_0_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_0_conv2d_4_out_channels)
+ self.module0_1 = Module0(conv2d_0_in_channels=module0_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_1_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_1_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_1_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_1_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_1_conv2d_4_out_channels)
+ self.module0_2 = Module0(conv2d_0_in_channels=module0_2_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_2_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_2_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_2_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_2_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_2_conv2d_4_out_channels)
+ self.module0_3 = Module0(conv2d_0_in_channels=module0_3_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_3_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_3_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_3_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_3_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_3_conv2d_4_out_channels)
+ self.module0_4 = Module0(conv2d_0_in_channels=module0_4_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_4_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_4_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_4_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_4_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_4_conv2d_4_out_channels)
+ self.module0_5 = Module0(conv2d_0_in_channels=module0_5_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_5_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_5_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_5_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_5_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_5_conv2d_4_out_channels)
+ self.module0_6 = Module0(conv2d_0_in_channels=module0_6_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_6_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_6_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_6_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_6_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_6_conv2d_4_out_channels)
+ self.module0_7 = Module0(conv2d_0_in_channels=module0_7_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_7_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_7_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_7_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_7_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_7_conv2d_4_out_channels)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ module0_1_opt = self.module0_1(module0_0_opt)
+ module0_2_opt = self.module0_2(module0_1_opt)
+ module0_3_opt = self.module0_3(module0_2_opt)
+ module0_4_opt = self.module0_4(module0_3_opt)
+ module0_5_opt = self.module0_5(module0_4_opt)
+ module0_6_opt = self.module0_6(module0_5_opt)
+ module0_7_opt = self.module0_7(module0_6_opt)
+ return module0_7_opt
+
+
+class Module12(nn.Cell):
+ def __init__(self, module0_0_conv2d_0_in_channels, module0_0_conv2d_0_out_channels, module0_0_conv2d_2_in_channels,
+ module0_0_conv2d_2_out_channels, module0_0_conv2d_4_in_channels, module0_0_conv2d_4_out_channels,
+ module0_1_conv2d_0_in_channels, module0_1_conv2d_0_out_channels, module0_1_conv2d_2_in_channels,
+ module0_1_conv2d_2_out_channels, module0_1_conv2d_4_in_channels, module0_1_conv2d_4_out_channels,
+ module0_2_conv2d_0_in_channels, module0_2_conv2d_0_out_channels, module0_2_conv2d_2_in_channels,
+ module0_2_conv2d_2_out_channels, module0_2_conv2d_4_in_channels, module0_2_conv2d_4_out_channels,
+ module0_3_conv2d_0_in_channels, module0_3_conv2d_0_out_channels, module0_3_conv2d_2_in_channels,
+ module0_3_conv2d_2_out_channels, module0_3_conv2d_4_in_channels, module0_3_conv2d_4_out_channels):
+ super(Module12, self).__init__()
+ self.module0_0 = Module0(conv2d_0_in_channels=module0_0_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_0_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_0_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_0_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_0_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_0_conv2d_4_out_channels)
+ self.module0_1 = Module0(conv2d_0_in_channels=module0_1_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_1_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_1_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_1_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_1_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_1_conv2d_4_out_channels)
+ self.module0_2 = Module0(conv2d_0_in_channels=module0_2_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_2_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_2_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_2_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_2_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_2_conv2d_4_out_channels)
+ self.module0_3 = Module0(conv2d_0_in_channels=module0_3_conv2d_0_in_channels,
+ conv2d_0_out_channels=module0_3_conv2d_0_out_channels,
+ conv2d_2_in_channels=module0_3_conv2d_2_in_channels,
+ conv2d_2_out_channels=module0_3_conv2d_2_out_channels,
+ conv2d_4_in_channels=module0_3_conv2d_4_in_channels,
+ conv2d_4_out_channels=module0_3_conv2d_4_out_channels)
+
+ def construct(self, x):
+ module0_0_opt = self.module0_0(x)
+ module0_1_opt = self.module0_1(module0_0_opt)
+ module0_2_opt = self.module0_2(module0_1_opt)
+ module0_3_opt = self.module0_3(module0_2_opt)
+ return module0_3_opt
+
+
+class MainModel(nn.Cell):
+ def __init__(self):
+ super(MainModel, self).__init__()
+ self.conv2d_0 = nn.Conv2d(in_channels=3,
+ out_channels=64,
+ kernel_size=(7, 7),
+ stride=(2, 2),
+ padding=(3, 3, 3, 3),
+ pad_mode="pad",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_1 = nn.ReLU()
+ self.pad_maxpool2d_2 = nn.Pad(paddings=((0, 0), (0, 0), (1, 0), (1, 0)))
+ self.maxpool2d_2 = nn.MaxPool2d(kernel_size=(3, 3), stride=(2, 2))
+ self.module8_0 = Module8(conv2d_0_in_channels=128,
+ conv2d_0_out_channels=256,
+ module3_0_conv2d_0_in_channels=64,
+ module3_0_conv2d_0_out_channels=128,
+ module3_0_conv2d_0_kernel_size=(1, 1),
+ module3_0_conv2d_0_stride=(1, 1),
+ module3_0_conv2d_0_padding=0,
+ module3_0_conv2d_0_pad_mode="valid",
+ module3_1_conv2d_0_in_channels=128,
+ module3_1_conv2d_0_out_channels=128,
+ module3_1_conv2d_0_kernel_size=(3, 3),
+ module3_1_conv2d_0_stride=(1, 1),
+ module3_1_conv2d_0_padding=(1, 1, 1, 1),
+ module3_1_conv2d_0_pad_mode="pad")
+ self.conv2d_4 = nn.Conv2d(in_channels=64,
+ out_channels=256,
+ kernel_size=(1, 1),
+ stride=(1, 1),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_10 = nn.ReLU()
+ self.module9_0 = Module9(module0_0_conv2d_0_in_channels=256,
+ module0_0_conv2d_0_out_channels=128,
+ module0_0_conv2d_2_in_channels=128,
+ module0_0_conv2d_2_out_channels=128,
+ module0_0_conv2d_4_in_channels=128,
+ module0_0_conv2d_4_out_channels=256,
+ module0_1_conv2d_0_in_channels=256,
+ module0_1_conv2d_0_out_channels=128,
+ module0_1_conv2d_2_in_channels=128,
+ module0_1_conv2d_2_out_channels=128,
+ module0_1_conv2d_4_in_channels=128,
+ module0_1_conv2d_4_out_channels=256)
+ self.module8_1 = Module8(conv2d_0_in_channels=256,
+ conv2d_0_out_channels=512,
+ module3_0_conv2d_0_in_channels=256,
+ module3_0_conv2d_0_out_channels=256,
+ module3_0_conv2d_0_kernel_size=(1, 1),
+ module3_0_conv2d_0_stride=(1, 1),
+ module3_0_conv2d_0_padding=0,
+ module3_0_conv2d_0_pad_mode="valid",
+ module3_1_conv2d_0_in_channels=256,
+ module3_1_conv2d_0_out_channels=256,
+ module3_1_conv2d_0_kernel_size=(3, 3),
+ module3_1_conv2d_0_stride=(2, 2),
+ module3_1_conv2d_0_padding=(1, 1, 1, 1),
+ module3_1_conv2d_0_pad_mode="pad")
+ self.conv2d_26 = nn.Conv2d(in_channels=256,
+ out_channels=512,
+ kernel_size=(1, 1),
+ stride=(2, 2),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_32 = nn.ReLU()
+ self.module9_1 = Module9(module0_0_conv2d_0_in_channels=512,
+ module0_0_conv2d_0_out_channels=256,
+ module0_0_conv2d_2_in_channels=256,
+ module0_0_conv2d_2_out_channels=256,
+ module0_0_conv2d_4_in_channels=256,
+ module0_0_conv2d_4_out_channels=512,
+ module0_1_conv2d_0_in_channels=512,
+ module0_1_conv2d_0_out_channels=256,
+ module0_1_conv2d_2_in_channels=256,
+ module0_1_conv2d_2_out_channels=256,
+ module0_1_conv2d_4_in_channels=256,
+ module0_1_conv2d_4_out_channels=512)
+ self.module0_0 = Module0(conv2d_0_in_channels=512,
+ conv2d_0_out_channels=256,
+ conv2d_2_in_channels=256,
+ conv2d_2_out_channels=256,
+ conv2d_4_in_channels=256,
+ conv2d_4_out_channels=512)
+ self.module8_2 = Module8(conv2d_0_in_channels=512,
+ conv2d_0_out_channels=1024,
+ module3_0_conv2d_0_in_channels=512,
+ module3_0_conv2d_0_out_channels=512,
+ module3_0_conv2d_0_kernel_size=(1, 1),
+ module3_0_conv2d_0_stride=(1, 1),
+ module3_0_conv2d_0_padding=0,
+ module3_0_conv2d_0_pad_mode="valid",
+ module3_1_conv2d_0_in_channels=512,
+ module3_1_conv2d_0_out_channels=512,
+ module3_1_conv2d_0_kernel_size=(3, 3),
+ module3_1_conv2d_0_stride=(2, 2),
+ module3_1_conv2d_0_padding=(1, 1, 1, 1),
+ module3_1_conv2d_0_pad_mode="pad")
+ self.conv2d_55 = nn.Conv2d(in_channels=512,
+ out_channels=1024,
+ kernel_size=(1, 1),
+ stride=(2, 2),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_61 = nn.ReLU()
+ self.module13_0 = Module13(module0_0_conv2d_0_in_channels=1024,
+ module0_0_conv2d_0_out_channels=512,
+ module0_0_conv2d_2_in_channels=512,
+ module0_0_conv2d_2_out_channels=512,
+ module0_0_conv2d_4_in_channels=512,
+ module0_0_conv2d_4_out_channels=1024,
+ module0_1_conv2d_0_in_channels=1024,
+ module0_1_conv2d_0_out_channels=512,
+ module0_1_conv2d_2_in_channels=512,
+ module0_1_conv2d_2_out_channels=512,
+ module0_1_conv2d_4_in_channels=512,
+ module0_1_conv2d_4_out_channels=1024,
+ module0_2_conv2d_0_in_channels=1024,
+ module0_2_conv2d_0_out_channels=512,
+ module0_2_conv2d_2_in_channels=512,
+ module0_2_conv2d_2_out_channels=512,
+ module0_2_conv2d_4_in_channels=512,
+ module0_2_conv2d_4_out_channels=1024,
+ module0_3_conv2d_0_in_channels=1024,
+ module0_3_conv2d_0_out_channels=512,
+ module0_3_conv2d_2_in_channels=512,
+ module0_3_conv2d_2_out_channels=512,
+ module0_3_conv2d_4_in_channels=512,
+ module0_3_conv2d_4_out_channels=1024,
+ module0_4_conv2d_0_in_channels=1024,
+ module0_4_conv2d_0_out_channels=512,
+ module0_4_conv2d_2_in_channels=512,
+ module0_4_conv2d_2_out_channels=512,
+ module0_4_conv2d_4_in_channels=512,
+ module0_4_conv2d_4_out_channels=1024,
+ module0_5_conv2d_0_in_channels=1024,
+ module0_5_conv2d_0_out_channels=512,
+ module0_5_conv2d_2_in_channels=512,
+ module0_5_conv2d_2_out_channels=512,
+ module0_5_conv2d_4_in_channels=512,
+ module0_5_conv2d_4_out_channels=1024,
+ module0_6_conv2d_0_in_channels=1024,
+ module0_6_conv2d_0_out_channels=512,
+ module0_6_conv2d_2_in_channels=512,
+ module0_6_conv2d_2_out_channels=512,
+ module0_6_conv2d_4_in_channels=512,
+ module0_6_conv2d_4_out_channels=1024,
+ module0_7_conv2d_0_in_channels=1024,
+ module0_7_conv2d_0_out_channels=512,
+ module0_7_conv2d_2_in_channels=512,
+ module0_7_conv2d_2_out_channels=512,
+ module0_7_conv2d_4_in_channels=512,
+ module0_7_conv2d_4_out_channels=1024)
+ self.module13_1 = Module13(module0_0_conv2d_0_in_channels=1024,
+ module0_0_conv2d_0_out_channels=512,
+ module0_0_conv2d_2_in_channels=512,
+ module0_0_conv2d_2_out_channels=512,
+ module0_0_conv2d_4_in_channels=512,
+ module0_0_conv2d_4_out_channels=1024,
+ module0_1_conv2d_0_in_channels=1024,
+ module0_1_conv2d_0_out_channels=512,
+ module0_1_conv2d_2_in_channels=512,
+ module0_1_conv2d_2_out_channels=512,
+ module0_1_conv2d_4_in_channels=512,
+ module0_1_conv2d_4_out_channels=1024,
+ module0_2_conv2d_0_in_channels=1024,
+ module0_2_conv2d_0_out_channels=512,
+ module0_2_conv2d_2_in_channels=512,
+ module0_2_conv2d_2_out_channels=512,
+ module0_2_conv2d_4_in_channels=512,
+ module0_2_conv2d_4_out_channels=1024,
+ module0_3_conv2d_0_in_channels=1024,
+ module0_3_conv2d_0_out_channels=512,
+ module0_3_conv2d_2_in_channels=512,
+ module0_3_conv2d_2_out_channels=512,
+ module0_3_conv2d_4_in_channels=512,
+ module0_3_conv2d_4_out_channels=1024,
+ module0_4_conv2d_0_in_channels=1024,
+ module0_4_conv2d_0_out_channels=512,
+ module0_4_conv2d_2_in_channels=512,
+ module0_4_conv2d_2_out_channels=512,
+ module0_4_conv2d_4_in_channels=512,
+ module0_4_conv2d_4_out_channels=1024,
+ module0_5_conv2d_0_in_channels=1024,
+ module0_5_conv2d_0_out_channels=512,
+ module0_5_conv2d_2_in_channels=512,
+ module0_5_conv2d_2_out_channels=512,
+ module0_5_conv2d_4_in_channels=512,
+ module0_5_conv2d_4_out_channels=1024,
+ module0_6_conv2d_0_in_channels=1024,
+ module0_6_conv2d_0_out_channels=512,
+ module0_6_conv2d_2_in_channels=512,
+ module0_6_conv2d_2_out_channels=512,
+ module0_6_conv2d_4_in_channels=512,
+ module0_6_conv2d_4_out_channels=1024,
+ module0_7_conv2d_0_in_channels=1024,
+ module0_7_conv2d_0_out_channels=512,
+ module0_7_conv2d_2_in_channels=512,
+ module0_7_conv2d_2_out_channels=512,
+ module0_7_conv2d_4_in_channels=512,
+ module0_7_conv2d_4_out_channels=1024)
+ self.module12_0 = Module12(module0_0_conv2d_0_in_channels=1024,
+ module0_0_conv2d_0_out_channels=512,
+ module0_0_conv2d_2_in_channels=512,
+ module0_0_conv2d_2_out_channels=512,
+ module0_0_conv2d_4_in_channels=512,
+ module0_0_conv2d_4_out_channels=1024,
+ module0_1_conv2d_0_in_channels=1024,
+ module0_1_conv2d_0_out_channels=512,
+ module0_1_conv2d_2_in_channels=512,
+ module0_1_conv2d_2_out_channels=512,
+ module0_1_conv2d_4_in_channels=512,
+ module0_1_conv2d_4_out_channels=1024,
+ module0_2_conv2d_0_in_channels=1024,
+ module0_2_conv2d_0_out_channels=512,
+ module0_2_conv2d_2_in_channels=512,
+ module0_2_conv2d_2_out_channels=512,
+ module0_2_conv2d_4_in_channels=512,
+ module0_2_conv2d_4_out_channels=1024,
+ module0_3_conv2d_0_in_channels=1024,
+ module0_3_conv2d_0_out_channels=512,
+ module0_3_conv2d_2_in_channels=512,
+ module0_3_conv2d_2_out_channels=512,
+ module0_3_conv2d_4_in_channels=512,
+ module0_3_conv2d_4_out_channels=1024)
+ self.module9_2 = Module9(module0_0_conv2d_0_in_channels=1024,
+ module0_0_conv2d_0_out_channels=512,
+ module0_0_conv2d_2_in_channels=512,
+ module0_0_conv2d_2_out_channels=512,
+ module0_0_conv2d_4_in_channels=512,
+ module0_0_conv2d_4_out_channels=1024,
+ module0_1_conv2d_0_in_channels=1024,
+ module0_1_conv2d_0_out_channels=512,
+ module0_1_conv2d_2_in_channels=512,
+ module0_1_conv2d_2_out_channels=512,
+ module0_1_conv2d_4_in_channels=512,
+ module0_1_conv2d_4_out_channels=1024)
+ self.module8_3 = Module8(conv2d_0_in_channels=1024,
+ conv2d_0_out_channels=2048,
+ module3_0_conv2d_0_in_channels=1024,
+ module3_0_conv2d_0_out_channels=1024,
+ module3_0_conv2d_0_kernel_size=(1, 1),
+ module3_0_conv2d_0_stride=(1, 1),
+ module3_0_conv2d_0_padding=0,
+ module3_0_conv2d_0_pad_mode="valid",
+ module3_1_conv2d_0_in_channels=1024,
+ module3_1_conv2d_0_out_channels=1024,
+ module3_1_conv2d_0_kernel_size=(3, 3),
+ module3_1_conv2d_0_stride=(2, 2),
+ module3_1_conv2d_0_padding=(1, 1, 1, 1),
+ module3_1_conv2d_0_pad_mode="pad")
+ self.conv2d_217 = nn.Conv2d(in_channels=1024,
+ out_channels=2048,
+ kernel_size=(1, 1),
+ stride=(2, 2),
+ padding=0,
+ pad_mode="valid",
+ dilation=(1, 1),
+ group=1,
+ has_bias=True)
+ self.relu_223 = nn.ReLU()
+ self.module9_3 = Module9(module0_0_conv2d_0_in_channels=2048,
+ module0_0_conv2d_0_out_channels=1024,
+ module0_0_conv2d_2_in_channels=1024,
+ module0_0_conv2d_2_out_channels=1024,
+ module0_0_conv2d_4_in_channels=1024,
+ module0_0_conv2d_4_out_channels=2048,
+ module0_1_conv2d_0_in_channels=2048,
+ module0_1_conv2d_0_out_channels=1024,
+ module0_1_conv2d_2_in_channels=1024,
+ module0_1_conv2d_2_out_channels=1024,
+ module0_1_conv2d_4_in_channels=1024,
+ module0_1_conv2d_4_out_channels=2048)
+ self.avgpool2d_238 = nn.AvgPool2d(kernel_size=(7, 7))
+ self.flatten_239 = nn.Flatten()
+ self.dense_240 = nn.Dense(in_channels=2048, out_channels=1000, has_bias=True)
+
+ def construct(self, input_1):
+ opt_conv2d_0 = self.conv2d_0(input_1)
+ opt_relu_1 = self.relu_1(opt_conv2d_0)
+ opt_maxpool2d_2 = self.pad_maxpool2d_2(opt_relu_1)
+ opt_maxpool2d_2 = self.maxpool2d_2(opt_maxpool2d_2)
+ module8_0_opt = self.module8_0(opt_maxpool2d_2)
+ opt_conv2d_4 = self.conv2d_4(opt_maxpool2d_2)
+ opt_add_9 = P.Add()(module8_0_opt, opt_conv2d_4)
+ opt_relu_10 = self.relu_10(opt_add_9)
+ module9_0_opt = self.module9_0(opt_relu_10)
+ module8_1_opt = self.module8_1(module9_0_opt)
+ opt_conv2d_26 = self.conv2d_26(module9_0_opt)
+ opt_add_31 = P.Add()(module8_1_opt, opt_conv2d_26)
+ opt_relu_32 = self.relu_32(opt_add_31)
+ module9_1_opt = self.module9_1(opt_relu_32)
+ module0_0_opt = self.module0_0(module9_1_opt)
+ module8_2_opt = self.module8_2(module0_0_opt)
+ opt_conv2d_55 = self.conv2d_55(module0_0_opt)
+ opt_add_60 = P.Add()(module8_2_opt, opt_conv2d_55)
+ opt_relu_61 = self.relu_61(opt_add_60)
+ module13_0_opt = self.module13_0(opt_relu_61)
+ module13_1_opt = self.module13_1(module13_0_opt)
+ module12_0_opt = self.module12_0(module13_1_opt)
+ module9_2_opt = self.module9_2(module12_0_opt)
+ module8_3_opt = self.module8_3(module9_2_opt)
+ opt_conv2d_217 = self.conv2d_217(module9_2_opt)
+ opt_add_222 = P.Add()(module8_3_opt, opt_conv2d_217)
+ opt_relu_223 = self.relu_223(opt_add_222)
+ module9_3_opt = self.module9_3(opt_relu_223)
+ opt_avgpool2d_238 = self.avgpool2d_238(module9_3_opt)
+ opt_flatten_239 = self.flatten_239(opt_avgpool2d_238)
+ opt_dense_240 = self.dense_240(opt_flatten_239)
+ return opt_dense_240