This is Barret Zoph's neural network machine translation and language modeling toolkit.
-This code allows for multi-GPU parallelization of LSTM machine translation and language models.
-The multi-GPU parallelization scheme is similar to that used in the Google paper: "Sequence to Sequence Learning with Neural Networks" (Link: http://papers.nips.cc/paper/5346-sequence-to-sequence-learning-with-neural-networks.pdf)
-For usage and instructions checkout the tutorial.pdf
-The code for the paper: "Multi-Source Neural Translation" by Zoph et al. (Link: http://arxiv.org/abs/1601.00710) is not currently available