This repository implemented various kinds of Neural Machine Translation models with Pytorch and TensorFlow, which is an open-source (MIT) neural machine translation system. It is designed to be research friendly to try out new ideas in translation.
There are 10k training data and 0.1k validation&test data in the file: ./data/
[News]
(19 Nov. 2018) // Adding basic seq2seq model (./seq2seq/seq2seq)
(21 Nov. 2018) // Adding seq2seq + attn model (./seq2seq/seq2seq_att)
(26 Nov. 2018) // Adding transformer-simple model (./transformer/transformer-pt-v1)
(26 Nov. 2018) // Adding transformer-simple tf-version (./transformer/transformer-tf-v1)
(01 Dev. 2018) // Adding GAN-NMT (./Adversarial-NMT)
(in progress) // Adding RGSE-NMT
(19 Jun. 2019) // Adding transformer model (./transformer/transformer-pt-v2)