The Replit x Weights & Biases Machine Learning Hackathon was Replit’s very first machine learning-focused hackathon that took place on February 4-11, 2023 with participants selected from all over the world!

After applying and being accepted into the hackathon, participants hacked for seven days on their projects, using both custom machine learning models and fine-tuning existing ones, all by combining the power of Weights & Biases and Replit. During the hackathon, builders collaborated and got help from mentors on our Discord (shout out to Morgan and Ayush!), worked through examples and demos from the opening ceremony and our list of resources, and even documented their journeys using W&B Reports.

In the end, we gave away over 500,000 Cycles worth of prizes to the best projects created by combining the two technologies. Expert ML engineers from Weights & Biases judged each project to decide the ultimate winners:

And the winners are...

Honorable Mention

We wanted to give a special shoutout to @NirantK for their proof of concept on how to use LLMs with Python to scale human effort. Using GPT-3, this demo fills in the correct model class and hyperparameters for a given script. Test it out and see if you’re convinced!

LLMs for WandB and Replit

Best Repl

Generate synthetic kōans on-demand with @vertinski’s Zen Master GPT project. Some highlights of this project were the GPT-2 model fine-tuned on a custom dataset, and the clean, calm interface. Try it out for yourself and get your Zen on!

Zen Master GPT Repl

Best W&B Report

@IcemasterEric implemented Q-Learning from scratch in this project, and the result was an engaging, day-by-day documentation of their journey. After in-depth research, pixel art and game design, and some debugging, the final product certainly caught our attention. After you give the full Report a read, you can play the game on Replit!

Q-Learning Repl

Grand Prize

This project combined the best features of both Weights & Biases and Replit – a robust Report and model, along with a polished, visually enthralling Repl. Without any previous ML experience, @codingMASTER398 used 40,000 images to train a Convolutional Neural Network on detecting specific filters in new images. For details on the process, check out the Report, and upload your own images to the Repl to find out what filters are detected.

Pancakes pic filter being detected

You can watch the closing ceremony on YouTube.

Thank you to everyone who participated, and thank you to Weights & Biases for sponsoring this hackathon! We can’t wait to see what else you build with machine learning on Replit.