Ellen Jiang

[Ellen Jiang @EllenJiang2]

What if research papers were easier to read? We came up with some ideas for using AI to enhance (not replace) the experience, then combined them to create Lumi!

Check out the blog post at https://t.co/Yy1JS0kDSy and final demo at https://t.co/BNqJlAqD8q 💜

[Ellen Jiang @EllenJiang2]

Belated update: We recently launched AutoNotes, an AI-notetaking experiment with open-source code + web demo now available!
@vivtsai @aconnors_writes @alemolina_ta @donsbach

https://t.co/ktaKYRmGc1

[Adam Pearce @adamrpearce]

Do Machine Learning Models Memorize or Generalize?

https://t.co/Ln3xIZhKLs

An interactive introduction to grokking and mechanistic interpretability w/ @ghandeharioun, @nadamused_, @Nithum, @wattenberg and @iislucas

[Adam Pearce @adamrpearce]

Search results that reflect historic inequities can amplify stereotypes and perpetuate under-representation. Carefully measuring diversity can help.

w/ @EllenJiang2 @dylnbkr @mmitchell_ai @timnitGebru

explorable: https://t.co/BAo2o6JPty
colab: https://t.co/622YVutHY8

[Adam Pearce @adamrpearce]

Why Some Models Leak Data: @EllenJiang2 and I visualized soccer and salary models to show how training data can be inadvertently revealed.

https://t.co/s4IO7DlHG7

[Adam Pearce @adamrpearce]

The availability of giant datasets and faster computers is making it harder to collect and study private information without inadvertently violating people’s privacy.

@EllenJiang2 and I took a look at how random numbers can help.

https://t.co/LLSRTlu91K

[Ian Tenney @iftenney]

Introducing an early-release version of the Language Interpretability Tool (LIT), a visual, interactive, and extensible open-source tool for analyzing all sorts of NLP models 🔥

Code: https://t.co/4CCWiAMPMk
Paper: https://t.co/2bkfx0C5lt
#NLProc

(1/4)

[Ellen Jiang @EllenJiang2]

How do machine learning models interpret paintings and optical illusions? Check out these blog posts I worked on with @emilyrreif and @_beenkim on depth in art history (https://t.co/QHhFW6e2D6) and depth model illusions (https://t.co/AXHvHAxufk).