🧵Can we reconcile excitement for SAEs with negative results? Our #ICML2026 position paper argues that even if SAEs underperform baselines when acting on knowns (e.g. probing, steering), they're a powerful tool for ~discovering unknowns~ Poster: Tue 2pm, HALL A #1715
Профиль
Kenny Peng
Профиль VivelyCS PhD student at Cornell Tech. Interested in interactions between algorithms and society. Princeton math '22. kennypeng.me
Presenting this at #COLT2026! Let me know if you want to chat about (1) the linear representation hypothesis, (2) theoretical foundations for interpretability / science of AI. - Talk: Thursday 2:33 pm - Deep Learning Theory / Optimization - Poster: Thursday Lunch)
Kenny PengNew paper! The Linear Representation Hypothesis is a powerful intuition for how language models work, but lacks formalization. We give a mathematical framework in which we can ask and answer a basic question: how many features can be stored under the hypothesis? 🧵 arxiv.org/abs/2602.11246
In college, I wrote this essay about Clark Kimberling, who maintains the Encyclopedia of Triangle Centers. The essay covers triangle geometry's rise (which includes Napoleon), its fall (from "brute force" methods), and its continued following ("the melody lingers on")
We made traversle.io, a new daily word game! The goal is to traverse from a start word to a target word through a network of related words. (Our motivating question: is it possible to construct a network that allows human navigation?)
Very cool — a way to jump from one post onto a trail of related posts. Lots of possibility and opportunity to push beyond the feed interface!
spacecowboyIntroducing a new way to explore Bluesky: foryou.club/also-liked Enter a post URL to see what other posts people who liked it also like. For example, enter the url of this art post: bsky.app/profile/alar... and you will get a bunch more art posts: foryou.club/also-liked?p...
🧵 On social media, we feel trapped in “filter bubbles” and “echo chambers.” Often, we see this as a problem with the algorithm. Our new essay argues the problem is the feed interface itself: it constrains our movement—all we can do is scroll. We lay out an alternative vision:
Excited to share our new research demo, where you can freely traverse the world of Bluesky through 20,000 interconnected trails, spanning “analysis of fictional tropes” to “rotisserie chicken” to “zoning and land use policy.” Try it out, and let us know what you think!