Joy of Computing

Language models built on datasets scraped from the open web have become foundational in natural language processing, but they reflect and amplify the biases and harms of their training data. We created a system which lets us filter training data to build more value-aligned models. It's imperfect—the internet is really big, harmful text is ever-evolving, and human labels include human biases. But little-by-little, we get closer to friendlier models! — mathemakitten

Jim is working on a vocoder at RC currently! You can read about it on his blog and check out the code here: https://github.com/jimkang/channel-vocoder. — Mai

A clever way to automatically format (and thus be able to see) all the cells in a Google Sheets spreadsheet that are intended to be inputs. — nicholasbs

Add tags to your starred repos on GitHub, so you never again lose track of that plugin or parser you totally meant to install. — rachel

Instead of suggesting books that are similar to ones you've already read, this recommendation engine tries to broaden your horizons by showing you what you're not reading or paying attention to. — Mai

A thoughtful blog post on the value of measuring vs. building. — Mai

Janet bindings to the J array programming language! — vaibhavsagar

Advice and suggestions for a set of tools for small teams to use for machine learning research projects. — nicholasbs

Create pixel art and music... with your (anonymized) face. — rachel

A Tweetcart is a game or visual demo that fits in 280 characters (or fewer) of PICO-8 code. Souren animated a wind sock, with 19 characters to spare! — rachel