caffe 运行过程中如何显示更新梯度信息 solver.cpp

在solver.cpp中。。。

包含文件变动如下：

#include <cstdio>


#include <algorithm>
#include <string>
#include <vector>
#include <iomanip>


#include "caffe/net.hpp"
#include "caffe/proto/caffe.pb.h"
#include "caffe/solver.hpp"
#include "caffe/util/io.hpp"
#include "caffe/util/math_functions.hpp"
#include "caffe/util/upgrade_proto.hpp"


using std::max;
using std::min;
using std::setw;
using std::scientific;
using std::left;

正文cpp文件中，void Solver<Dtype>::Solve(const char* resume_file) {

中变动如下：

if (display) {
      LOG(INFO) << "Iteration " << iter_ << ", loss = " << loss;
      const vector<Blob<Dtype>*>& result = net_->output_blobs();
      int score_index = 0;
      for (int j = 0; j < result.size(); ++j) {
        const Dtype* result_vec = result[j]->cpu_data();
        const string& output_name =
            net_->blob_names()[net_->output_blob_indices()[j]];
        const Dtype loss_weight =
            net_->blob_loss_weights()[net_->output_blob_indices()[j]];
        for (int k = 0; k < result[j]->count(); ++k) {
          ostringstream loss_msg_stream;
          if (loss_weight) {
            loss_msg_stream << " (* " << loss_weight
                            << " = " << loss_weight * result_vec[k] << " loss)";
          }
          LOG(INFO) << "    Train net output #"
              << score_index++ << ": " << output_name << " = "
              << result_vec[k] << loss_msg_stream.str();
        }
      }

    //add by me
     int max_len = 0;
      for (int l = 0; l < net_->layers().size(); ++l) {
        Layer<Dtype>& layer = *net_->layers()[l].get();
        if (layer.blobs().size() > 0 && layer.layer_param().name().length()
            > max_len)
          max_len = layer.layer_param().name().length();
      }
     for (int l = 0; l < net_->layers().size(); ++l) {
        Layer<Dtype>& layer = *net_->layers()[l].get();
        for (int b = 0; b < layer.blobs().size(); ++b) {
          Blob<Dtype>& blob = *layer.blobs()[b].get();
          const Dtype* blob_cpu_data = blob.cpu_data();
          const Dtype* blob_cpu_diff = blob.cpu_diff();
          Dtype data_sum = 0;
          Dtype diff_sum = 0;
          for (int i = 0; i < blob.count(); ++i) {
            data_sum += (blob_cpu_data[i] > Dtype(0.)) ? blob_cpu_data[i]
                : - blob_cpu_data[i];
            diff_sum += (blob_cpu_diff[i] > Dtype(0.)) ? blob_cpu_diff[i]
                : - blob_cpu_diff[i];
          }
          data_sum /= blob.count();
          diff_sum /= blob.count();
          LOG(INFO) << std::left << std::setw(max_len + 1) << std::setfill(' ')
                    << layer.layer_param().name()
                    << " blob" << b << ": " << std::scientific
                   << data_sum << " [" << diff_sum << "]";
        }
      }

     //end revised by me

}