Ben Ambridge

ambridge.bsky.social

18 October 2026 12:40

arxiv.org/pdf/2610.17538 “Everyone Was Right: Lexical Representations in Language Models Are Both Abstract and Holistic” - just as I argued (for kids!) in doi.org/10.1177/0142...

David Sidhu

davidmsidhu.bsky.social

17 October 2026 16:53

🚨 New Paper Alert! I look at new predictors of iconicity to learn more about what KINDS of words are iconic. tl;dr Iconic English words are more familiar, have more meanings, and are more emotional. They are also more specific when abstract (but more on this below). doi.org/10.1016/j.co...

Caroline Rowland

carorowland.bsky.social

17 October 2026 12:29

📢 Two initiatives for displaced students and academics that deserve more publicity: 🇩🇪 www.maxminds.mpg.de - Max Planck mentorship scheme for displaced scholars 🇳🇱 www.dutchscholarsforpalestine.nl/scholarships... scholarships for Palestinians in NL If you know of more please add in reply

William Timkey

wtimkey.bsky.social

16 October 2026 22:56

Transformer LMs develop attention heads that appear to track syntactic dependencies. But do these heads actually drive syntax-sensitive predictions? In a new preprint with @tallinzen.bsky.social, we put this to the test. The answer was surprisingly mixed. 🧵 Paper link: arxiv.org/abs/2610.192...

Gary Lupyan

glupyan.bsky.social

15 October 2026 21:14

Do LLMs reason by pattern matching? Why, yes they do! But as we argue here, so do people! @zstuddiford.bsky.social started this work 4 years ago when he was a freshman in my lab. Zach is now at UCSD w/ Ben Bergen and I couldn't be more proud of him and of this project. Have a look!