AlexanderFabisch/deep_learning_installation.md

## deep_learning_installation.md

      
    Raw
  

              deep_learning_installation.md
            
          
    My system


Ubuntu 14.04
GeFore GTX970

Table of Contents


Install GPU Driver
Install CUDA
Install cuDNN
Install Caffe
Install Theano
Install Tensorflow
Install Keras

Install GPU Driver


must be an official Nvidia driver version >=352
you should install it from the Ubuntu packages: sudo apt-get install nvidia-352
otherwise, get it from http://www.nvidia.com/download/driverResults.aspx/98373/en-us and run

sudo service lightdm stop
sudo sh NVIDIA-Linux-x86_64-361.28.run
Install CUDA

Download it from https://developer.nvidia.com/cuda-downloads -> Linux -> x86_64 -> Ubuntu -> 14.04 -> deb (local)
instructions:
sudo dpkg -i cuda-repo-ubuntu1404-7-5-local_7.5-18_amd64.deb
sudo apt-get update
sudo apt-get install cuda
sudo reboot
add to .bashrc:
export PATH=/usr/local/cuda-<version>/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-<version>/lib64:$LD_LIBRARY_PATH
try to run the example:
cuda-install-samples-7.5.sh ~  # copies examples to ~/NVIDIA_CUDA-Samples_<version>
cd ~/NVIDIA_CUDA-Samples_7.5/5_Simulations/nbody
make
./nbody
Install cuDNN

download it from https://developer.nvidia.com/rdp/cudnn-download, requirement: CUDA compute capabilities >= 3.5
instructions:
tar -xzvf cudnn-7.0-linux-x64-v4.0-prod.tgz
cd cuda
sudo cp lib64/* /usr/local/cuda/lib64
sudo cp include/cudnn.h /usr/local/cuda/include
Install Caffe

instructions:

install dependencies

sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev libsnappy-dev
# install leveldb:
wget https://leveldb.googlecode.com/files/leveldb-1.9.0.tar.gz
tar -xzf leveldb-1.9.0.tar.gz
cd leveldb-1.9.0
make
sudo mv libleveldb.* /usr/local/lib
cd include
sudo cp -R leveldb /usr/local/include
sudo ldconfig

Configure the build by copying and modifying the example Makefile.config
for your setup. The defaults should work, but uncomment the relevant lines
if using Anaconda Python.

cp Makefile.config.example Makefile.config
# Adjust Makefile.config (for example, if using Anaconda Python, or if cuDNN is desired)
make all && make test && make runtest

For cuDNN acceleration using NVIDIA’s proprietary cuDNN software,
uncomment the USE_CUDNN := 1 switch in Makefile.config. cuDNN is
ometimes but not always faster than Caffe’s GPU acceleration.
To compile the Python and MATLAB wrappers do make pycaffe and
make matcaffe respectively. Be sure to set your MATLAB and Python
paths in Makefile.config first!

Install Theano

instructions:
sudo apt-get install python-pip python-dev libopenblas-dev
sudo pip install numpy scipy
git clone git://github.com/Theano/Theano.git
cd Theano
python setup.py develop --user
You can run a simple test program to see whether Theano works with your GPU. See here for details.
Install Tensorflow

requirement: cuDNN for the GPU version
instructions
sudo pip install --upgrade https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow-0.8.0rc0-cp27-none-linux_x86_64.whl
Install Keras

instructions

install Theano and/or Tensorflow

git clone https://github.com/fchollet/keras.git
cd keras
sudo python setup.py install
# make the config directory accessible!
sudo chown -R <username> ~/.keras
Switching between backends:
The configuration is stored in ~/.keras/keras.json:
{"epsilon": 1e-07, "floatx": "float32", "backend": "tensorflow"}
"backend" can either be "tensorflow" or "theano".