Skip to content

Latest commit

 

History

History
8 lines (6 loc) · 427 Bytes

README.md

File metadata and controls

8 lines (6 loc) · 427 Bytes

seq2seq learning

  • this is motivated by TensorFlow's tutorial: https://www.tensorflow.org/tutorials/seq2seq
  • but the model structure is more like that stated in this paper: https://arxiv.org/abs/1409.0473
  • I wrote these codes to help me better understanding the details of the seq2seq model
  • WARNING: these are very rush codes,
  • Although I've got reasonable results (in terms of BLEU) on the 12 million en-fr dataset