7、caffe之train函数片段之初始网络开始

当运行下列命令的时候

ubuntu@ubuntu:~/caffe$ ./examples/mnist/train_lenet.sh 

这是脚本train_lenet.sh 命令行（如果只有cpu 需要修改这个文件的lenet_solver.prototxt，选择 cpu）

#!/usr/bin/env sh
set -e

./build/tools/caffe train --solver=examples/mnist/lenet_solver.prototxt $@

下面是caffe.cpp里面的函数段

// Train / Finetune a model.
int train() {
  CHECK_GT(FLAGS_solver.size(), 0) << "Need a solver definition to train.";
  CHECK(!FLAGS_snapshot.size() || !FLAGS_weights.size())
      << "Give a snapshot to resume training or weights to finetune "
      "but not both.";
  vector<string> stages = get_stages_from_flags();

  caffe::SolverParameter solver_param;
  caffe::ReadSolverParamsFromTextFileOrDie(FLAGS_solver, &solver_param);

  solver_param.mutable_train_state()->set_level(FLAGS_level);
  for (int i = 0; i < stages.size(); i++) {
    solver_param.mutable_train_state()->add_stage(stages[i]);
  }

  // If the gpus flag is not provided, allow the mode and device to be set
  // in the solver prototxt.
  if (FLAGS_gpu.size() == 0
      && solver_param.has_solver_mode()
      && solver_param.solver_mode() == caffe::SolverParameter_SolverMode_GPU) {
      if (solver_param.has_device_id()) {
          FLAGS_gpu = "" +
              boost::lexical_cast<string>(solver_param.device_id());
      } else {  // Set default GPU if unspecified
          FLAGS_gpu = "" + boost::lexical_cast<string>(0);
      }
  }

  vector<int> gpus;
  get_gpus(&gpus);
  if (gpus.size() == 0) {
    LOG(INFO) << "Use CPU.";
    Caffe::set_mode(Caffe::CPU);
  } else {
    ostringstream s;
    for (int i = 0; i < gpus.size(); ++i) {
      s << (i ? ", " : "") << gpus[i];
    }
    LOG(INFO) << "Using GPUs " << s.str();
#ifndef CPU_ONLY
    cudaDeviceProp device_prop;
    for (int i = 0; i < gpus.size(); ++i) {
      cudaGetDeviceProperties(&device_prop, gpus[i]);
      LOG(INFO) << "GPU " << gpus[i] << ": " << device_prop.name;
    }
#endif
    solver_param.set_device_id(gpus[0]);
    Caffe::SetDevice(gpus[0]);
    Caffe::set_mode(Caffe::GPU);
    Caffe::set_solver_count(gpus.size());
  }

  caffe::SignalHandler signal_handler(
        GetRequestedAction(FLAGS_sigint_effect),
        GetRequestedAction(FLAGS_sighup_effect));

  shared_ptr<caffe::Solver<float> >
      solver(caffe::SolverRegistry<float>::CreateSolver(solver_param));

  solver->SetActionFunction(signal_handler.GetActionFunction());

  if (FLAGS_snapshot.size()) {
    LOG(INFO) << "Resuming from " << FLAGS_snapshot;
    solver->Restore(FLAGS_snapshot.c_str());
  } else if (FLAGS_weights.size()) {
    CopyLayers(solver.get(), FLAGS_weights);
  }

  LOG(INFO) << "Starting Optimization";
  if (gpus.size() > 1) {
#ifdef USE_NCCL
    caffe::NCCL<float> nccl(solver);
    nccl.Run(gpus, FLAGS_snapshot.size() > 0 ? FLAGS_snapshot.c_str() : NULL);
#else
    LOG(FATAL) << "Multi-GPU execution not available - rebuild with USE_NCCL";
#endif
  } else {
    solver->Solve();
  }
  LOG(INFO) << "Optimization Done.";
  return 0;
}
RegisterBrewFunction(train);

--solver=examples/mnist/lenet_solver.prototxt 文件里面的参数

代码中如FLAGS_solver等都是命令解析的参数gflags的参数使用
关于train test time 注册 前面已经写过demo 不在详细解释

vector< string> stages = get_stages_from_flags();

这函数主要是针对命令行多个参数进行统计（详细参见本博客c++基本知识讲解）

 caffe::SolverParameter solver_param;

这行代码表示使用了probuff解析了caffe.proto文件，然后使用的变量在caffe.pb.cc文件里面定义了

caffe::ReadSolverParamsFromTextFileOrDie(FLAGS_solver, &solver_param);

函数的原定义在/caffe/src/caffe/util/update_proto.cpp中，这里主要是把命令行中​​--solver=examples/mnist/lenet_solver.prototxt​​​的lenet_solver.prototxt里面的参数读到caffe.proto变量中；
过程：FLAGS_solver 存放着命令行solver=“路径”；然后调用util/io.hpp文件中的函数进行ReadProtoFromTextFile（…)读操作 ；之后又调用io.cpp文件中的ReadProtoFromTextFile（…）函数，并写入了caffe.proto生成的.cc文件中；

shared_ptr<caffe::Solver<float> >
solver(caffe::SolverRegistry<float>::CreateSolver(solver_param));

进入了caffe/include/solver_factory.hpp文件,进行”注册
简单总结如下：
(1) 在caffe中，不同的配置文件中存在不同的梯度下降算法；以minist 手写为例，因为其lenet_solver.prototxt 配置文件中并未指示明确，但是在caffe.proto中，已经默认设置梯度下降算法SGD（其它可参考不同的配置文件：如lenet_adadelta_solver.prototxt；lenet_solver_adam.prototxt）
SolverRegistry::CreateSolver(solver_param)。
（2）然后匹配到map中注册过字符串和函数指针，匹配到了 SGDSolver；
（3）执行new SGDSolver(solver_param)创建solver。
（4） 又因为 new 的新对象 SGDSolver ，需要构造自身的构造函数，因为其是子类，因此需要先初始化父类solver.cpp的构造函数；所以执行了slover.cpp 构造函数，进行了初始化网络
具体流程：
程序启动之前：以sgd函数为例 其它函数类似（AdaDelta，RMSProp，…）
执行sgd_solver.cpp 宏定义REGISTER_SOLVER_CLASS(SGD);
然后执行solver_factory.hpp 宏展开#define REGISTER_SOLVER_CLASS(type) 的 REGISTER_SOLVER_CREATOR(type, Creator_##type##Solver)的各类型梯度函数注册，第二个参数为

Solver* Creator_##type##Solver( 
 const SolverParameter& param) 
 { LOG(INFO) << "000000 solver type: " ; 
 return new type##Solver(param); 
 }
 然后定义触发
 #define REGISTER_SOLVER_CREATOR(type, creator) 
 static SolverRegisterer g_creator_f_##type(#type, creator); 
 static SolverRegisterer g_creator_d_##type(#type, creator) 
 在然后进行注册：
 SolverRegistry::AddCreator(type, creator);
 完成注册之后；
 caffe.cpp 的
 shared_ptr<caffe::Solver >
 solver(caffe::SolverRegistry::CreateSolver(solver_param));
 会进入solver_factory.hpp 执行registry[type](param);进行函数回调（map）；
 将具体执行
 Solver* Creator_##type##Solver( 
 const SolverParameter& param) 
 { LOG(INFO) << "000000 solver type: " ; 
 return new type##Solver(param); 
 }

子类继承父类构造 --初始化网络

   Solver<Dtype>::Solver(const SolverParameter& param)
    : net_(), callbacks_(), requested_early_exit_(false) {
  Init(param);
}

可以进入ubuntu系统的文件/tmp目录，会在运行一次产生caffe.ubuntu.ubuntu.log.INFO.20170627-141243.4440 日志信息，可以跟踪其打印的日志信息跟踪代码的阅读
下面是截取一小段glogs日志信息

Log file created at: 2017/06/27 14:12:43
Running on machine: ubuntu
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
I0627 14:12:43.187585  4440 caffe.cpp:211] Use CPU.
I0627 14:12:43.188031  4440 solver.cpp:44] Initializing solver from parameters: 
test_iter: 100
test_interval: 500
base_lr: 0.01
display: 100
max_iter: 10000
lr_policy: "inv"
gamma: 0.0001
power: 0.75
momentum: 0.9
weight_decay: 0.0005
snapshot: 5000
snapshot_prefix: "examples/mnist/lenet"
solver_mode: CPU
net: "examples/mnist/lenet_train_test.prototxt"
train_state {
  level: 0
  stage: ""
}
I0627 14:12:43.188195  4440 solver.cpp:87] Creating training net from net file: examples/mnist/lenet_train_test.prototxt
I0627 14:12:43.188627  4440 net.cpp:294] The NetState phase (0) differed from the phase (1) specified by a rule in layer mnist
I0627 14:12:43.188666  4440 net.cpp:294] The NetState phase (0) differed from the phase (1) specified by a rule in layer accuracy
I0627 14:12:43.188832  4440 net.cpp:51] Initializing net from parameters: 
name: "LeNet"
state {
  phase: TRAIN
  level: 0
  stage: ""
}
layer {