
Sophia Sanborn
@naturecomputes
Searching for Principles of Neural Representation | Neuro + AI @ enigmaproject.ai | @stanford | @neur_reps | 🦋 @naturecomputes.bsky.social
ID: 971175623539359745
http://sophiasanborn.com 07-03-2018 00:08:07
721 Tweet
5,5K Followers
2,2K Following

🎉 Good news! The fun doesn’t stop after the first social! Join us for a joint party with Symmetry and Geometry in Neural Representations and NeuroAI Workshop @ NeurIPS2024 . Don’t miss out—scan the QR code to register now! 🚀


We are excited to hold our post-workshop social jointly with UniReps, NeuroAI Workshop @ NeurIPS2024, enigmaproject.ai and New Theory. To register, scan the QR code!


“Maths has been an incredible tool for describing physics. In the same way I think AI might be an incredible descriptive language for biology.” - Demis Hassabis, awarded the 2024 Nobel Prize in Chemistry for his work using artificial intelligence to predict the 3D structure of



Evgenii Egorov Yeah; the underlying question seems whether you consider graphs to be geometric objects or not. I'd argue opinions are divided :) This great survey (Sophia Sanborn Mathilde Papillon🦋 mathildepapillon .bsky .social Christian Shewmake Abby Bertics Nina Miolane 🦋 @ninamiolane.bsky.social et al) calls them 'topological': arxiv.org/abs/2407.09468


Episode #24 in #TheoreticalNeurosciencePodcast: On neuroscience foundation models - with Andreas Tolias Andreas Tolias Lab @ Stanford University theoreticalneuroscience.no/thn24 The Tolias lab is building generally applicable “foundation models” for optophysiological responses in mouse visual cortex.




New NanoGPT training speed world record from the Enigma Project 🎉 (Andreas Tolias Lab @ Stanford University, Sophia Sanborn, enigmaproject.ai) We improve the efficiency of gradient all_reduce. Short explainer of our method 👇 [1/6]

🚀 Congrats to Konstantin Willeke and the Enigma Project team enigmaproject.ai for being NanoGPT speed‑run world record-champions for a glorious day 🏆🔥


