Ann Huang

annhuang42.bsky.social

14 October 2026 19:21

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)

Mohadeseh Shafiei Kafraj

mohishafiei.bsky.social

27 September 2026 23:55

I’ll be presenting a new biologically plausible associative memory model tomorrow at the Bernstein Conference, in the Hippocampal Memory Circuits workshop. Supervised by Peter Latham, in collaboration with @krotov.bsky.social @apalmigiano.bsky.social @saxelab.bsky.social @yiotapoirazi.bsky.social.

Yoav Ger

yoavger.bsky.social

25 September 2026 13:19

Excited to share that our work, "Learning Reveals Invisible Structure in Low-Rank RNNs", has been selected for an 𝗢𝗿𝗮𝗹 presentation (top 1%) at #NeurIPS 2026! 🎉 Very grateful for this special recognition, and even more grateful to my incredible co-author and mentor Omri Barak.

Francis Bach

bachfrancis.bsky.social

25 September 2026 13:19

Overwhelmed by the variance of means of exponentials? This blog post is meant for you. As a bonus, a spectral closed-form formula for (approximate) multinomial logistic regression. francisbach.com/spectral_log...

Andrew Lampinen

lampinen.bsky.social

23 September 2026 20:35

New post reflecting on recent AI progress in math, how AI is changing the way we work, and some worries about where people will find meaning as they offload more of their work to AI: infinitefaculty.substack.com/p/math-resea...