caffe跑mnist数据集

1.下载数据集：可以通过脚本下载也可以直接到官网下载了放到data/mnist下，然后解压

2.设置用cpu训练

lenet_solver.prototxt:

# solver mode: CPU or GPU
solver_mode: CPU

create_mnist.sh中的数据集名字一定要和data/mnist下数据的名字一样，否则会保打不开文件的错误

cd $CAFFE_ROOT

./examples/mnist/create_mnist.sh

cd $CAFFE_ROOT
./examples/mnist/train_lenet.sh

注意：遇到的问题及解决

1.

F1124 12:36:09.104321 18800 convert_mnist_data.cpp:48] Check failed: image_file Unable to open file data/mnist/t10k-images-idx3-ubyte
*** Check failure stack trace: ***
    @     0x7f16b54eddaa  (unknown)
    @     0x7f16b54edce4  (unknown)
    @     0x7f16b54ed6e6  (unknown)
    @     0x7f16b54f0687  (unknown)
    @           0x40393a  convert_dataset()
    @           0x40242d  main
    @     0x7f16b46fdec5  (unknown)
    @           0x402476  (unknown)
    @              (nil)  (unknown)
Aborted (core dumped)
原因：数据集名字和create_mnist.sh中的一样，重命名(下载下来的数据名字中有个地方是.引用的时候用的-所以要改为一样的)

2.

./create_mnist.sh: 16: ./create_mnist.sh: build/examples/mnist/convert_mnist_data.bin: not found