A Fullstack web application that serves as an experiment tracking tool, which allows users to tune a Deep Learning model that solves the MNIST challenge as a black box.
A record of my developement journey, everything from progress, bugs, to struggles and breakthroughs. It's mostly for me to reflect on everyday and treasure how far I have come. If you are remotely interested, definitely check out me talking to myself :)
- Cuong Nguyen
- Anh Phan Github link
- Simple form: A form to submit job configurations, implemented with input validity checks.
- Message board: A notification interface to help users understand the current process. Message content includes confirmation of job submission and report of submitted jobs (both in queue and finished).
- Progress bar: A progress bar that shows users training progress in real time!
- Job board: A table of 10 best job configurations that gets updated everytime a training + evaluation cycle finishes.
- Frontend: React.js, Bootstrap, React Bootstrap
- Backend: Python, Flask, Flask-SocketIO, pika, CORS, MongoEngine, PyTorch, Threading
- Database: MongoDB
- JobQueue: RabbitMQ
- Hosing service: Netlify, Render, CloudAMQP
- Misc (Things I used but unsure how to classify): Docker, Postman, Werkzeug, gunicorn, waitress, eventlet, gevent, kombu