Keras implementation for seq2seq
In this project I try to implement seq2seq word level model using keras.
Resource Used:
- MSRP paraphrase corpus
Requirements:
- Keras
- Numpy
- The official tutorial about implementing seq2seq model is character level. But word level implementation is common.
python basic_seq2seq.py
test_source.txt
original sentencetest_target.txt
translation sentencetrain_vocab.txt
vocabulary
- The inference function has some problems
- Write the evaluation code
- Add Attention function
- Using Bidirectional LSTM in encoder
- Implementing Joint Copying and Restricted Generation for Paraphrase