I built a simple chatbot using conversations from Cornell University's Movie Dialogue Corpus. The main features of our model are LSTM cells, a bidirectional dynamic RNN, and decoders with attention.
If you would like to learn a bit more about the details of this project, especially the sequence-to-sequence portion, I wrote an article about this.
movie_conversations.txt.zip contains the indexes for the lines of each conversation. movie_lines.txt.zip contains the lines spoken by the characters.
To view my work most easily, click on the .ipynb file.