基于C++接口实现端侧训练

查看源文件

注意:MindSpore已经统一端边云推理API,如您想继续使用MindSpore Lite独立API进行端侧训练,可以参考此文档

概述

本教程基于LeNet训练示例代码,演示在Android设备上训练一个LeNet。

端侧训练流程如下:

  1. 基于MindSpore构建训练模型,并导出MindIR模型文件。

  2. 使用MindSpore Lite Converter工具,将MindIR模型转为端侧MS模型。

  3. 调用MindSpore Lite训练API,加载端侧MS模型,执行训练。

下面章节首先通过示例代码中集成好的脚本,帮你快速部署并执行示例,再详细讲解实现细节。

准备

推荐使用Ubuntu 18.04 64位操作系统。

环境要求

  • 系统环境:Linux x86_64,推荐使用Ubuntu 18.04.02LTS

  • 软件依赖

    • GCC >= 7.3.0

    • CMake >= 3.18.3

    • Git >= 2.28.0

    • Android_NDK >= r20

      • 配置环境变量:export ANDROID_NDK=NDK路径

下载数据集

示例中的MNIST数据集由10类28*28的灰度图片组成,训练数据集包含60000张图片,测试数据集包含10000张图片。

MNIST数据集官网下载地址:http://yann.lecun.com/exdb/mnist/,共4个下载链接,分别是训练数据、训练标签、测试数据和测试标签。

下载并解压到本地,解压后的训练和测试集分别存放于/PATH/MNIST_Data/train/PATH/MNIST_Data/test路径下。

目录结构如下:

MNIST_Data/
├── test
│   ├── t10k-images-idx3-ubyte
│   └── t10k-labels-idx1-ubyte
└── train
    ├── train-images-idx3-ubyte
    └── train-labels-idx1-ubyte

安装MindSpore

你可以通过pip或是源码的方式安装MindSpore,详见MindSpore官网安装教程

下载并安装MindSpore Lite

通过git克隆源码,进入源码目录,Linux指令如下:

git clone https://gitee.com/mindspore/mindspore.git -b r1.8
cd ./mindspore

源码路径下的mindspore/lite/examples/train_lenet_cpp目录包含了本示例程序的源码。其中version和下文中MindSpore Lite下载页面的version保持一致。如果-b 指定master,需要通过源码编译的方式获取对应的安装包。

请到MindSpore Lite下载页面下载mindspore-lite-{version}-linux-x64.tar.gz以及mindspore-lite-{version}-android-aarch64.tar.gz。其中,mindspore-lite-{version}-linux-x64.tar.gz是MindSpore Lite在x86平台的安装包,里面包含模型转换工具converter_lite,本示例用它来将MINDIR模型转换成MindSpore Lite支持的.ms格式;mindspore-lite-{version}-android-aarch64.tar.gz是MindSpore Lite在Android平台的安装包,里面包含训练运行时库libmindspore-lite.so,本示例用它所提供的接口在Android上训练模型。最后将文件放到MindSpore源码下的output目录(如果没有output目录,请创建它)。

假设下载的安装包存放在/Downloads目录,上述操作对应的Linux指令如下:

mkdir output
cp /Downloads/mindspore-lite-{version}-linux-x64.tar.gz output/mindspore-lite-{version}-linux-x64.tar.gz
cp /Downloads/mindspore-lite-{version}-android-aarch64.tar.gz output/mindspore-lite-{version}-android-aarch64.tar.gz

您也可以通过源码编译直接生成端侧训练框架对应的x86平台安装包mindspore-lite-{version}-linux-x64.tar.gz以及Android平台安装包mindspore-lite-{version}-android-aarch64.tar.gz,源码编译的安装包会自动生成在output目录下,请确保output目录下同时存在这两个安装包。

连接安卓设备

准备好一台Android设备,并通过USB与工作电脑正确连接。手机需开启“USB调试模式”,华为手机一般在设置->系统和更新->开发人员选项->USB调试中打开“USB调试模式”。

本示例使用adb工具与Android设备进行通信,在工作电脑上远程操控移动设备;如果没有安装adb工具,可以执行apt install adb安装。

模型训练和验证

进入示例代码目录并执行训练脚本,Linux指令如下:

cd mindspore/lite/examples/train_lenet_cpp
bash prepare_and_run.sh -D /PATH/MNIST_Data -t arm64

其中/PATH/MNIST_Data是你工作电脑上存放MNIST数据集的绝对路径,-t arm64为执行训练和推理的设备类型,如果工作电脑连接多台手机设备,可使用-i devices_id指定运行设备。

prepare_and_run.sh脚本做了以下工作:

  1. 导出lenet_tod.mindir模型文件;

  2. 调用上节的模型转换工具将lenet_tod.mindir转换为lenet_tod.ms文件;

  3. lenet_tod.ms、MNIST数据集和相关依赖库文件推送至你的Android设备;

  4. 执行训练、保存并推理模型。

Android设备上训练LeNet模型每轮会输出损失值和准确率;最后选择训练完成的模型执行推理,验证MNIST手写字识别精度。端侧训练LeNet模型10个epoch的结果如下所示(测试准确率会受设备差异的影响):

======Training Locally=========
1.100:  Loss is 1.19449
1.200:  Loss is 0.477986
1.300:  Loss is 0.440362
1.400:  Loss is 0.165605
1.500:  Loss is 0.368853
1.600:  Loss is 0.179764
1.700:  Loss is 0.173386
1.800:  Loss is 0.0767713
1.900:  Loss is 0.493
1.1000: Loss is 0.460352
1.1100: Loss is 0.262044
1.1200: Loss is 0.222022
1.1300: Loss is 0.058006
1.1400: Loss is 0.0794117
1.1500: Loss is 0.0241433
1.1600: Loss is 0.127109
1.1700: Loss is 0.0557566
1.1800: Loss is 0.0698758
Epoch (1):      Loss is 0.384778
Epoch (1):      Training Accuracy is 0.8702
2.100:  Loss is 0.0538642
2.200:  Loss is 0.444504
2.300:  Loss is 0.0806976
2.400:  Loss is 0.0495807
2.500:  Loss is 0.178903
2.600:  Loss is 0.265705
2.700:  Loss is 0.0933796
2.800:  Loss is 0.0880472
2.900:  Loss is 0.0480734
2.1000: Loss is 0.241272
2.1100: Loss is 0.0920451
2.1200: Loss is 0.371406
2.1300: Loss is 0.0365746
2.1400: Loss is 0.0784372
2.1500: Loss is 0.207537
2.1600: Loss is 0.442626
2.1700: Loss is 0.0814725
2.1800: Loss is 0.12081
Epoch (2):      Loss is 0.176118
Epoch (2):      Training Accuracy is 0.94415
......
10.1000:        Loss is 0.0984653
10.1100:        Loss is 0.189702
10.1200:        Loss is 0.0896037
10.1300:        Loss is 0.0138191
10.1400:        Loss is 0.0152357
10.1500:        Loss is 0.12785
10.1600:        Loss is 0.026495
10.1700:        Loss is 0.436495
10.1800:        Loss is 0.157564
Epoch (5):     Loss is 0.102652
Epoch (5):     Training Accuracy is 0.96805
AvgRunTime: 18980.5 ms
Total allocation: 125829120
Accuracy is 0.965244

===Evaluating trained Model=====
Total allocation: 20971520
Accuracy is 0.965244

===Running Inference Model=====
There are 1 input tensors with sizes:
tensor 0: shape is [32 32 32 1]
There are 1 output tensors with sizes:
tensor 0: shape is [32 10]
The predicted classes are:
4, 0, 2, 8, 9, 4, 5, 6, 3, 5, 2, 1, 4, 6, 8, 0, 5, 7, 3, 5, 8, 3, 4, 1, 9, 8, 7, 3, 0, 2, 3, 6,

如果你没有Android设备,也可以执行bash prepare_and_run.sh -D /PATH/MNIST_Data -t x86直接在PC上运行本示例。

示例程序详解

示例程序结构

  train_lenet_cpp/
  ├── model
  │   ├── lenet_export.py
  │   ├── prepare_model.sh
  │   └── train_utils.py
  |
  ├── scripts
  │   ├── batch_of32.dat
  │   ├── eval.sh
  │   ├── infer.sh
  │   └── train.sh
  │
  ├── src
  │   ├── inference.cc
  │   ├── net_runner.cc
  │   ├── net_runner.h
  │   └── utils.h
  │
  ├── Makefile
  ├── README.md
  ├── README_CN.md
  └── prepare_and_run.sh

定义并导出模型

首先我们需要基于MindSpore框架创建一个LeNet模型,本例中直接用MindSpore ModelZoo的现有LeNet模型

本小结使用MindSpore云侧功能导出,更多信息请参考MindSpore教程

import numpy as np
import mindspore as ms
from lenet import LeNet5
from train_utils import TrainWrap

n = LeNet5()
n.set_train()
ms.set_context(mode=ms.GRAPH_MODE, device_target="CPU", save_graphs=False)

然后定义输入和标签张量大小:

BATCH_SIZE = 32
x = ms.Tensor(np.ones((BATCH_SIZE, 1, 32, 32)), ms.float32)
label = ms.Tensor(np.zeros([BATCH_SIZE]).astype(np.int32))
net = TrainWrap(n)

定义损失函数、网络可训练参数、优化器,并启用单步训练,由TrainWrap函数实现。

from mindspore import nn
import mindspore as ms

def train_wrap(net, loss_fn=None, optimizer=None, weights=None):
    """
    train_wrap
    """
    if loss_fn is None:
        loss_fn = nn.SoftmaxCrossEntropyWithLogits(reduction='mean', sparse=True)
    loss_net = nn.WithLossCell(net, loss_fn)
    loss_net.set_train()
    if weights is None:
        weights = ms.ParameterTuple(net.trainable_params())
    if optimizer is None:
        optimizer = nn.Adam(weights, learning_rate=0.003, beta1=0.9, beta2=0.999, eps=1e-5, use_locking=False, use_nesterov=False, weight_decay=4e-5, loss_scale=1.0)
    train_net = nn.TrainOneStepCell(loss_net, optimizer)
    return train_net

最后调用export接口将模型导出为MindIR文件保存(目前端侧训练仅支持MindIR格式)。

ms.export(net, x, label, file_name="lenet_tod", file_format='MINDIR')
print("finished exporting")

如果输出finished exporting表示导出成功,生成的lenet_tod.mindir文件在../train_lenet_cpp/model目录下。完整代码参见lenet_export.pytrain_utils.py

转换模型

prepare_model.sh中使用MindSpore Lite converter_lite工具将lenet_tod.mindir转换为ms模型文件,执行指令如下:

./converter_lite --fmk=MINDIR --trainModel=true --modelFile=lenet_tod.mindir --outputFile=lenet_tod

转换成功后,当前目录下会生成lenet_tod.ms模型文件。

更多用法参见训练模型转换

训练模型

模型训练的处理详细流程请参考net_runner.cc源码

模型训练的主函数为:

int NetRunner::Main() {
  // Load model and create session
  InitAndFigureInputs();
  // initialize the dataset
  InitDB();
  // Execute the training
  TrainLoop();
  // Evaluate the trained model
  CalculateAccuracy();

  if (epochs_ > 0) {
    auto trained_fn = ms_file_.substr(0, ms_file_.find_last_of('.')) + "_trained.ms";
    mindspore::Serialization::ExportModel(*model_, mindspore::kMindIR, trained_fn, mindspore::kNoQuant, false);
    trained_fn = ms_file_.substr(0, ms_file_.find_last_of('.')) + "_infer.ms";
    mindspore::Serialization::ExportModel(*model_, mindspore::kMindIR, trained_fn, mindspore::kNoQuant, true);
  }
  return 0;
}
  1. 加载模型

    InitAndFigureInputs函数加载转换后的MS模型文件,调用Graph接口创建graph_实例(下述代码中的ms_file_就是转换模型阶段生成的lenet_tod.ms模型)。

    void NetRunner::InitAndFigureInputs() {
      auto context = std::make_shared<mindspore::Context>();
      auto cpu_context = std::make_shared<mindspore::CPUDeviceInfo>();
      cpu_context->SetEnableFP16(enable_fp16_);
      context->MutableDeviceInfo().push_back(cpu_context);
    
      graph_ = new mindspore::Graph();
      auto status = mindspore::Serialization::Load(ms_file_, mindspore::kMindIR, graph_);
      if (status != mindspore::kSuccess) {
        std::cout << "Error " << status << " during serialization of graph " << ms_file_;
        MS_ASSERT(status != mindspore::kSuccess);
      }
    
      auto cfg = std::make_shared<mindspore::TrainCfg>();
      if (enable_fp16_) {
        cfg.get()->optimization_level_ = mindspore::kO2;
      }
    
      model_ = new mindspore::Model();
      status = model_->Build(mindspore::GraphCell(*graph_), context, cfg);
      if (status != mindspore::kSuccess) {
        std::cout << "Error " << status << " during build of model " << ms_file_;
        MS_ASSERT(status != mindspore::kSuccess);
      }
    
      acc_metrics_ = std::shared_ptr<AccuracyMetrics>(new AccuracyMetrics);
      model_->InitMetrics({acc_metrics_.get()});
    
      auto inputs = model_->GetInputs();
      MS_ASSERT(inputs.size() >= 1);
      auto nhwc_input_dims = inputs.at(0).Shape();
    
      batch_size_ = nhwc_input_dims.at(0);
      h_ = nhwc_input_dims.at(1);
      w_ = nhwc_input_dims.at(2);
    }
    
  2. 数据集处理

    InitDB函数预处理MNIST数据集并加载至内存。MindData提供了数据预处理API,用户可参见C++ API 说明文档 获取更多详细信息。

    int NetRunner::InitDB() {
      train_ds_ = Mnist(data_dir_ + "/train", "all", std::make_shared<SequentialSampler>(0, 0));
    
      TypeCast typecast_f(mindspore::DataType::kNumberTypeFloat32);
      Resize resize({h_, w_});
      train_ds_ = train_ds_->Map({&resize, &typecast_f}, {"image"});
    
      TypeCast typecast(mindspore::DataType::kNumberTypeInt32);
      train_ds_ = train_ds_->Map({&typecast}, {"label"});
    
      train_ds_ = train_ds_->Batch(batch_size_, true);
    
      if (verbose_) {
        std::cout << "DatasetSize is " << train_ds_->GetDatasetSize() << std::endl;
      }
      if (train_ds_->GetDatasetSize() == 0) {
        std::cout << "No relevant data was found in " << data_dir_ << std::endl;
        MS_ASSERT(train_ds_->GetDatasetSize() != 0);
      }
      return 0;
    }
    
  3. 执行训练

    首先创建训练回调类对象(例如LRSchedulerLossMonitorTrainAccuracyCkptSaver)数组指针;然后调用TrainLoop类的Train函数,将模型设置为训练模式;最后在训练过程中遍历执行回调类对象对应的函数并输出训练日志。CkptSaver会根据设定训练步长数值为当前会话保存CheckPoint模型,CheckPoint模型包含已更新的权重,在应用崩溃或设备出现故障时可以直接加载CheckPoint模型,继续开始训练。

    int NetRunner::TrainLoop() {
      mindspore::LossMonitor lm(100);
      mindspore::TrainAccuracy am(1);
    
      mindspore::CkptSaver cs(kSaveEpochs, std::string("lenet"));
      Rescaler rescale(kScalePoint);
      Measurement measure(epochs_);
    
      if (virtual_batch_ > 0) {
        model_->Train(epochs_, train_ds_, {&rescale, &lm, &cs, &measure});
      } else {
        struct mindspore::StepLRLambda step_lr_lambda(1, kGammaFactor);
        mindspore::LRScheduler step_lr_sched(mindspore::StepLRLambda, static_cast<void *>(&step_lr_lambda), 1);
        model_->Train(epochs_, train_ds_, {&rescale, &lm, &cs, &am, &step_lr_sched, &measure});
      }
    
      return 0;
    }
    
  4. 验证精度

    训练结束后调用CalculateAccuracy评估模型精度。该函数调用AccuracyMetricsEval方法,将模型设置为推理模式。

    float NetRunner::CalculateAccuracy(int max_tests) {
      test_ds_ = Mnist(data_dir_ + "/test", "all");
      TypeCast typecast_f(mindspore::DataType::kNumberTypeFloat32);
      Resize resize({h_, w_});
      test_ds_ = test_ds_->Map({&resize, &typecast_f}, {"image"});
    
      TypeCast typecast(mindspore::DataType::kNumberTypeInt32);
      test_ds_ = test_ds_->Map({&typecast}, {"label"});
      test_ds_ = test_ds_->Batch(batch_size_, true);
    
      model_->Evaluate(test_ds_, {});
      std::cout << "Accuracy is " << acc_metrics_->Eval() << std::endl;
    
      return 0.0;
    }