Amazon AWS上Caffe+GPU CUDA 8.0 +cuDNN 5.0 +OpenBLAS+pycaffe配置和lenet训练方法

//环境：Amazon AWS g2.2xlarge实例，Ubuntu 16.04, python2.7, Nvidia cuda 8,
cuDNN 5.0, OpenBLAS

sudo apt-get update

sudo apt-get install -y python-pip python-numpy python-scipy python-matplotlib

sudo apt-get install -y libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler

sudo apt-get install -y --no-install-recommends libboost-all-dev

sudo apt-get install -y libopenblas-dev

sudo apt-get install -y libgflags-dev libgoogle-glog-dev liblmdb-dev

sudo apt-get install -y unzip cmake

wget https://github.com/BVLC/caffe/archive/master.zip
unzip master.zip

cd caffe-master

cp Makefile.config.example Makefile.config

//安装NVIDIA CUDA Toolkit 8.0

cd ~

wget http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_8.0.61-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1604_8.0.61-1_amd64.deb

rm -rf cuda-repo-ubuntu1604_8.0.61-1_amd64.deb

sudo apt-get update

sudo apt-get install -y cuda

//安装NVIDIA cuDNN库

sudo apt-get install -y lrzsz
https://developer.nvidia.com/rdp/cudnn-download 注册下载 cudnn-8.0-linux-x64-v5.0.tgz

tar -xzvf cudnn-8.0-linux-x64-v5.0-ga.tgz

rm -rf cudnn-8.0-linux-x64-v5.0-ga.tgz

sudo cp -R cuda/lib64/lib* /usr/local/cuda/lib64/

sudo cp cuda/include/cudnn.h /usr/local/cuda/include

#编译安装caffe, pycaffe 

# Adjust Makefile.config (for example, if using Anaconda Python, or if cuDNN is desired)

cd ~/caffe-master

mkdir build

cd build

cmake -DCPU_ONLY=off -DBLAS=Open ..

make all -j8

make test

make runtest

cd ~/caffe-master/python

for req in $(cat requirements.txt); do pip install $req; done

export PYTHONPATH=~/caffe-master/python:$PYTHONPATH

#测试pycaffe安装是否正确

python

>>import caffe

#训练lenet

cd data/mnist

./get_mnist.sh

cd ../../

./examples/mnist/create_mnist.sh

./examples/mnist/train_lenet.sh

./build/tools/caffe test \

-model examples/mnist/lenet_train_test.prototxt \

-weights examples/mnist/lenet_iter_10000.caffemodel \

-iterations 100

 
#神经网络可视化

/root/caffe-master/python

sudo apt-get install graphviz

pip install pydot

pip install -r requirements.txt
python draw_net.py ../models/bvlc_reference_caffenet/train_val.prototxt caffenet.png

参考：

1. https://github.com/BVLC/caffe/pull/1667
2. http://www.cnblogs.com/zjutzz/p/6083201.html ILSVRC 2012图像下载。

3. https://www.zybuluo.com/nrailgun/note/488084 数据集

4. http://baike.baidu.com/item/%E6%B5%8B%E8%AF%95%E9%9B%86 测试集的名词解释

5.《深度学习 21天实战Caffe》