poncho

A Deep Learning seq2seq chatbot.

Initial steps

You will first need to fullfill some dependencies. So, run python -r requirements.txt in poncho and in nmt_chatbot. We recommend using vitualenv or similar environments to make things a bit easy to manage.

Steps to clean and prepare the data

This repo comes with multiple files to clean and prepare your input data. But fear not, we have a clean CLI to help you perform the cleanup steps that you require. We have used Reddit comment dumps from files.pushshift.io. So any data that is similarly structured will work.

For instructions on available options, simply run python manage.py -h in poncho directory.

Steps to train the chatbot

1. Place your input text in nmt_chatbot/new_data and name the file as train.from
2. Then place the expected output in nmt-chatbot/new_data and name it as train.to
3. You can leave the provided test data as it is.
4. Navigate to nmt_chatbot/settings.py and modify the settings as per your wish. The settings provided by default should be able to get a very basic chatbot running. We would recommend increasing the vocabulary size before modifying any other setting. Documentation for nmt_chatbot can be found here.
5. Next step is to prepare the input data. Navigate to nmt_chatbot/setup and run prepare_data.py. This generates the vocabulary based on the settings provided.
6. Now, to train the chatbot. Go to nmt_chatbot and run train.py.
7. Boom! You now have a chatbot that's ready to roll.

Seeing the chatbot in action

1. To run the chatbot, navigate to gui and run fileuploader.py and server.py in two separate terminals/command prompts
2. Now open up any browser of your choice and type in localhost:8888. Now you can interact with the chatbot!

---

Example screenshots