Interactive BPE Tokenizer - Capstone 2024 @ UCT

Final UCT CS project, 96% final mark.

Our Interactive BPE Tokenizer serves as a hands-on educational tool to help interested individuals learn the fundamentals of Natural Language Processing (NLP) by training and testing BPE tokenizers.

Members

• Ethan Lawrence (LWRETH001)
• Kirsten Sutherland (STHKIR002)
• Conor Karl McKeag (MCKCON007)

USAGE

Set ENV values

Create a .env file in /frontend

Add: VITE_BACKEND_API_URL="http://127.0.0.1:5000"

Docker Instructions

Ensure you have Docker/Docker Desktop installed.

docker-compose up -d

It may take some time but it will eventually be running in the background.

Frontend

After docker is done building the frontend should be available at http://localhost:8080

Docs

The docs are available at either this GitPage, or locally at http://localhost:3000

MongoDB

It is running at localhost:27017

Username: capstone2024

Password: changeme123

Make sure you have the mongoDB extension installed, MongoDB for VS Code