A Simple ChatBot Based Encoder-Decoder
This is a implementation of simplified and modular version of pytorch CHATBOT TUTORIAL
Data
datasets/conversations.csv contain formatted and preprocessed conversations from Cornell Movie-Dialogs Corpus.
train model
python3 main.py
You can configure encoder-decoder model in main.py
Evaluate model
Modify model_path variable in evaluate.py into your trained model path
Run python3 evaluate.py to chat with your trained model
output example:
Load checkpint...
Input> hello?
Chat_bot> hello .
Input> where am I?
Chat_bot> you re going to bed .
Input> how are you doing?
Chat_bot> fine .
Input> are you my friend?
Chat_bot> yes .
Input> you're under arrest
Chat_bot> you re not here .
Input> i'm just kidding
Chat_bot> no you re not .
Input> where are you from?
Chat_bot> i m here .
Input> it's time for me to leave
Chat_bot> i m sorry .