《caffe模型——prototxt表示》

~/caffe$ ./build/tools/caffe train –solver=examples/mnist/lenet_lr_solver.prototxt

这里值得注意的是：src/caffe/layer.cpp 直接用安装caffe时的源文件，不用更改。
src/caffe/layer.cpp :

include “caffe/layer.hpp”

namespace caffe {
INSTANTIATE_CLASS(Layer);
} // namespace caffe

终端输入：~/caffe$ ./build/tools/caffe train –solver=examples/mnist/lenet_lr_solver.prototxt

I0913 10:54:49.716719 8761 solver.cpp:414] Test net output #0: accuracy = 0.9223
I0913 10:54:49.716747 8761 solver.cpp:414] Test net output #1: loss = 0.281053 (* 1 = 0.281053 loss)
I0913 10:54:49.716754 8761 solver.cpp:332] Optimization Done.
I0913 10:54:49.716764 8761 caffe.cpp:250] Optimization Done.

注：solver_mode:GPU CPU 获得的准确度一样
若有其他错误，可能在Makefile.config