A machine translation system using LSTM attention model
================== this model is based on dynet neural network library which is implemented by Carnegie Mellon University and many other developers. the low-layer code is written by using C++, low-layer functions are developed based on Eigen, which is a C++ template library for linear algebra: matrices, vectors, numerical solvers, and related algorithms.
not only they develop the C++ interface which is usually called cnn, but also they implement the python interface pycnn, and dynet is actually similar with pycnn.