Ann Huang

annhuang42.bsky.social

14 October 2026 16:04

📄 Paper: arxiv.org/abs/2609.38081 🌐 Page: ann-huang-0.github.io/Hessian-null... 💻 Code: github.com/Ann-Huang-0/hessian-null-space-continuation w/ my awesome team 🌟 @neurostrow.bsky.social, @hansonlu.bsky.social @wtredman.bsky.social, @leokoz8.bsky.social, @kanakarajanphd.bsky.social (12/12)

Ann Huang

annhuang42.bsky.social

14 October 2026 16:03

How much representational diversity exists around a trained network? A lot, it turns out! Moving a ViT’s weights by 1.3% while preserving its predictions, its representations end up less similar to the original than a random-init ViT’s ‼️ New paper: Hessian Null Space Continuation (HNC) 💫🧵 (1/12)

Kanishka Misra

kanishka.bsky.social

14 October 2026 16:03

I think the 2004 Elman paper (and its 2008 follow-up) might be my all time fav Elman papers!

Tom McCoy

rtommccoy.bsky.social

14 October 2026 16:03

Properties that are enabled by being quasi-symbolic rather than precisely symbolic include: 1. Gradient similarity 2. Graceful degradation 3. Softness of constraints 4. Flexibility in learning 5. Handling of partially-symbolic tasks 7/n

Tom McCoy

rtommccoy.bsky.social

14 October 2026 16:02

🤖🧠NEW PAPER🧠🤖 What is the relationship between neural networks & symbolic systems? Based on evidence from mechanistic interpretability, we argue for a quasi-symbolic view of cognition: Neural nets implement symbols but only approximately Link: www.sciencedirect.com/science/arti... 1/n