Basic Chatbot Trained on Reddit Data

This code takes suriyadeepan/practical_seq2seq and applies it to a subset of the may reddit dataset.

First, clone the repo

git clone https://github.com/Data-Bot-Box/SnarkyChatBot2.git

Next, download reddit shower thoughts dataset from here Then, run the following code (from ~/SnarkyChatBot2/datasets/may_reddit):

pip3 install nltk
python3 data.py

After the data is extracted, run the following code to go back to the source directory and make a checkpoint directory and train the model.

cd ../../
mkdir ckpt
mkdir ckpt/reddit
python3 train_chatbot.py

The model should train from there. The code will also run inference on the test set. This inference code can be used to integrate with the chatbot.

timgasper/SnarkyChatBot2