Skip to content

Latest commit

 

History

History
162 lines (126 loc) · 4.69 KB

INSTALL.md

File metadata and controls

162 lines (126 loc) · 4.69 KB
Raw

Installing the pyCNN module.

(for instructions on installing on a computer with GPU, see below)

First, get CNN:

cd $HOME
git clone https://github.com/clab/cnn.git
cd cnn
git submodule init # To be consistent with CNN's installation instructions.
git submodule update # To be consistent with CNN's installation instructions.

Then get Eigen:

cd $HOME
cd cnn
# Latest version (17.03.16) of Eigen fails to compile , so we revert "-r" to the latest stable version.
# Otherwise, we can use "hg clone https://bitbucket.org/eigen/eigen/"
hg clone https://bitbucket.org/eigen/eigen/ -r 47fa289dda2dc13e0eea70adfc8671e93627d466

To simplify the following steps, we can set a bash variable to hold where we have saved the main directories of cnn and eigen. In case you have gotten ccn and eigen differently from the instructions above and saved them in different location(s), these variables will be helpful:

PATH_TO_CNN=$HOME/cnn/
PATH_TO_EIGEN=$HOME/cnn/eigen/

Compile CNN. (modify the code below to point to the correct boost location)

cd $PATH_TO_CNN
mkdir build
cd build
cmake .. -DEIGEN3_INCLUDE_DIR=$PATH_TO_EIGEN -DBOOST_ROOT=$HOME/.local/boost_1_58_0 -DBoost_NO_BOOST_CMAKE=ON
make -j 2

If CNN fails to compile and throws an error like this:

$ make -j 2
Scanning dependencies of target cnn
Scanning dependencies of target cnn_shared
[  1%] [  2%] Building CXX object cnn/CMakeFiles/cnn.dir/cfsm-builder.cc.o
Building CXX object cnn/CMakeFiles/cnn_shared.dir/cfsm-builder.cc.o
In file included from /home/user/cnn/cnn/cnn.h:13:0,
                 from /home/user/cnn/cnn/cfsm-builder.h:6,
                 from /home/user/cnn/cnn/cfsm-builder.cc:1:
/home/user/cnn/cnn/tensor.h:22:42: fatal error: unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include <unsupported/Eigen/CXX11/Tensor>
                                          ^
compilation terminated.

If CNN fails to compile with the error above, then you can download a stable version of Eigen and re-build CNN as such:

cd $PATH_TO_CNN
wget u.cs.biu.ac.il/~yogo/eigen.tgz
tar zxvf eigen.tgz # or "dtrx eigen.tgz" if you have dtrx installed
mkdir build
cd build
cmake .. -DEIGEN3_INCLUDE_DIR=$PATH_TO_EIGEN -DBOOST_ROOT=$HOME/.local/boost_1_58_0 -DBoost_NO_BOOST_CMAKE=ON
make -j 2

Now that CNN is compiled, we need to compile the pycnn module. This requires having cython installed. If you don't have cython, it can be installed with either pip install cython or better yet conda install cython.

pip2 install cython --user

Customize the setup.py to include (i) the parent directory where the main cnn directory is saved and (ii) the path to the main eigen directy:

cd $PATH_TO_CNN/pycnn
sed -i  "s|..\/..\/cnn\/|$PATH_TO_CNN|g" setup.py 
sed -i  "s|..\/..\/eigen\/|$PATH_TO_EIGEN|g" setup.py
make
make install

We are almost there. We need to tell the environment where to find the compiled cnn shared library. The pyCNN's make fetched a copy of libcnn_shared.so and put it in the pycnn lib.

Add the following line to your profile (.zshrc or .bashrc), change according to your installation location.

export LD_LIBRARY_PATH=$PATH_TO_CNN/pycnn

Now, check that everything works:

# check that it works:
cd $PATH_TO_CNN
cd pyexamples
python2 xor.py
python2 rnnlm.py rnnlm.py

Alternatively, if the following script works for you, then your installation is likely to be working:

from pycnn import *
model = Model()

Installing with GPU support

For installing on a computer with GPU, first install CUDA. Here, we assume CUDA is installed in /usr/local/cuda-7.5

There are two modules, pycnn which is the regular CPU module, and gpycnn which is the GPU module. You can import either of them, these are two independent modules. The GPU support is incomplete: some operations (i.e. hubber_distance) are not available for the GPU.

First step is to build the CNN modules. Checkout and go to the build directory (same instructions as above). Then:

To build a CPU version on a computer with CUDA:

cmake .. -DEIGEN3_INCLUDE_DIR=../eigen -DBACKEND=eigen
make -j 4

To build a GPU version on a computer with CUDA:

cmake .. -DBACKEND=cuda -DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-7.5/
make -j 4

Now, build the python modules (as above, we assume cython is installed):

The GPU module (gpycnn):

cd ../pycnn
make gpycnn.so
make ginstall

The CPU module (pycnn):

cd ../pycnn
make pycnn.so
make install

Add the following to your env: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PATH_TO_CNN/pycnn

Once both the pycnn and gpycnn are installed, run python ../pyexamples/cpu_vs_gpu.py for a small timing example.