langulski / chatbot-in-python

🤖 Chat Bot using Pytorch | Neural nets

Project Status

✔️ Complete

Table of Contents

• Objective
• Process
• Results
• Learning Process
• Authors

Objective

The objective of this project is to build a simple chatbot and deploy it on a web-page using "Flask". The bot was created using Pytorch and exemplifies how a neural net works. The project overall is very simple and provides an understanding of these subjects.

Process

• Create a JSON file with the intends;
• Transform these words to a language the machine can understand;
• Apply neural net;
• Create a train file and check if the results are valid;
• Add the chatbot to a webpage using Flask as backend;

Results

After training the bot and implementing Flask:

We named the bot as "Helo" and she was trained to understand questions asked in Portuguese:

Learning Process

Theory Applied

• Numpy
• Pytorch
• Neural Nets
• Flask

Challenges

• Training the bot to understand the words was challenging, especially because Portuguese have some special characters like (^~`´).

Improvements

• The JSON file does not have much info, we could improve this in order to get better responses from the bot.

Authors

Lucas Angulski

sources: https://chatbotsmagazine.com/contextual-chat-bots-with-tensorflow-4391749d0077

Huge thanks to Python Engineer for providing an awesome Pytorch tutorial.