In the repository, I'll look into the performance of some famous deep learning methods for NLP tesk. Thus, I adopt the quickest implement of Seq2seq Chatbot by Tensorflow's new rnn API and show the excellent result. Hope you and me will get some insight into Seq2seq after completing the notebook.

For more instruction and demonstration of Basic Seq2seq, please read the chatbot_test.ipynb.

A better Chinese instruction of this project can be found at zhihu: https://zhuanlan.zhihu.com/p/34357774

Q: do you have a free moment ?
A: sure . what do you need ?
G: yes , i have a few days .

Q: it happened at my house .
A: was anything stolen ?
G: i ' m sorry , sir . i ' ll have to get you in

Q: who is he marring ?
A: a girl he met on holiday in spain , i think .
G: he ' s a very good guy .

Tensorflow's new Seq2seq API tutorial: https://www.tensorflow.org/tutorials/seq2seq

A Tensorflow's legacy_seq2seq Chatbot: https://github.com/nicolas-ivanov/tf_seq2seq_chatbot

Adversarial-Learning-for-Neural-Dialogue-Generation-in-Tensorflow https://github.com/liuyuemaicha/Adversarial-Learning-for-Neural-Dialogue-Generation-in-Tensorflow