Sign in

Daria Lioubashevski

@darialioub.bsky.social
20 followers 53 following 21 posts

PhD student @ Huji | Student Researcher @Google Interested in the overlap of mech interp & cog comp neuroscience daria-lioubashevski.github.io

PostsRepliesMedia
Daria Lioubashevski @darialioub.bsky.social · 15/04/2026
We are excited to share our new preprint! 🚨 biorxiv.org/content/10.1... In this work, we investigate the neural mechanisms by which speakers generate information-rich linguistic content during spontaneous natural conversation. 🧵
biorxiv.org
100
Daria Lioubashevski @darialioub.bsky.social · 24/03/2026
Pre-print of a super cool project by the Hasson Lab 👶🏻
000
Daria Lioubashevski @darialioub.bsky.social · 21/12/2025
Beyond our main results, we explore a broader theoretical question in cognitive neuroscience: How are words represented in the brain? 🧠 LLMs turn out to be a useful lens for this long-standing debate 👇
130
Reposted by Daria Lioubashevski
Nature Computational Science @natcomputsci.nature.com · 18/11/2025
📢 @samnastase.bsky.social and colleagues show that aligning ECoG data into a shared space improves how well LLMs predict brain activity during language comprehension. www.nature.com/articles/s43... #compneuro #Neuroscience #ArtificialIntelligence 🔓 rdcu.be/eQzYm
nature.com
Aligning brains into a shared space improves their alignment with large language models - Nature Computational Science
Aligning electrocorticography data into a shared space improves how large language models predict brain activity during language comprehension, enhancing encoding accuracy, cross-participant generaliz...
184
Daria Lioubashevski @darialioub.bsky.social · 11/11/2025
🚨 New preprint! One idea, many ways to say it – but does your brain track those options while you speak? Using LLMs, we put this to the test. www.biorxiv.org/content/10.1... We show for the 1st time that the brain represents multiple alternatives simultaneously in both listening and speaking. 🧵
1155
Reposted by Daria Lioubashevski
Grace @gracekind.net · 05/10/2025
Update: @vgel.me has written a blog post about why this happens from a mechanistic interpretability standpoint! vgel.me/posts/seahor...
vgel.me
Why do LLMs freak out over the seahorse emoji?
Investigating the seahorse emoji doom loop using logitlens.
38812
Reposted by Daria Lioubashevski
Subha Nawer Pushpita @pushpita1729.bsky.social · 07/11/2025
Excited to share our work on mechanisms of naturalistic audiovisual processing in the human brain 🧠🎬!! www.biorxiv.org/content/10.1...
biorxiv.org
965
Reposted by Daria Lioubashevski
Victoria Bosch @initself.bsky.social · 03/11/2025
Introducing CorText: a framework that fuses brain data directly into a large language model, allowing for interactive neural readout using natural language. tl;dr: you can now chat with a brain scan 🧠💬 1/n
714253