Mantidae is a successor of Mantis (https://github.com/trevorcohn/mantis), but with more functionalities. With Mantidae, you can build a complete neural machine translation system with ease. Mantidae works fast and its performance is competitive with lamtram (https://github.com/neubig/lamtram) or nematus (https://github.com/rsennrich/nematus).

Before compiling Mantidae, you need:

• Eigen, using the development version (not release), e.g. 3.3.beta2 (http://bitbucket.org/eigen/eigen/get/3.3-beta2.tar.bz2)

• cuda version 7.5 or higher

• boost, e.g., 1.58 using libboost-all-dev ubuntu package

• cmake, e.g., 3.5.1 using cmake ubuntu package

• dynet. I myself modify some functions of dynet. So, dynet will be integrated inside Mantidae. You don't need to install it.

First, clone the repository

git clone https://github.com/duyvuleo/Mantidae.git

As mentioned above, you'll need the latest version of eigen (3.3.x or higher)

hg clone https://bitbucket.org/eigen/eigen/

A modified version of dynet (https://github.com/clab/dynet/tree/master/dynet) is already included (e.g., dynet folder).

Compiling to execute on a CPU is as follows

mkdir build_cpu
cd build_cpu
cmake .. -DEIGEN3_INCLUDE_DIR=eigen [-DBoost_NO_BOOST_CMAKE=ON]
make -j 2

Boost note. The "-DBoost_NO_BOOST_CMAKE=ON" can be optional but if you have a trouble of boost-related build error(s), adding it will help to overcome.

MKL support. If you have Intel's MKL library installed on your machine, you can speed up the computation on the CPU by:

cmake .. -DEIGEN3_INCLUDE_DIR=EIGEN [-DBoost_NO_BOOST_CMAKE=ON] -DMKL=TRUE -DMKL_ROOT=MKL

substituting in different paths to EIGEN and MKL if you have placed them in different directories.

This will build the 3 binaries

build_cpu/src/attentional
build_cpu/src/biattentional
build_cpu/src/relopt-decoder

Building on the GPU uses the Nvidia CUDA library, currently tested against version 7.5. The process is as follows

mkdir build_gpu
cd build_gpu
cmake .. -DBACKEND=cuda -DEIGEN3_INCLUDE_DIR=EIGEN -DCUDA_TOOLKIT_ROOT_DIR=CUDA [-DBoost_NO_BOOST_CMAKE=ON]
make -j 2

substituting in your Eigen and CUDA folders, as appropriate.

This will result in some binaries, including:

build_gpu/src/attentional
build_gpu/src/biattentional
build_gpu/src/relopt-decoder

(to be updated)

Hoang Cong Duy Vu ([email protected] or [email protected]), Trevor Cohn and Reza Haffari

Updated Mar 2017