[Reading Group]: Information Locality as an Inductive Bias for Neural Language Models

[nilinykh]

📆 Date & Time

Friday 11th of September, 15:00-17:00

📍 Location or Link

CSE EDIT 8103

https://chalmers.zoom.us/j/5240594528?omn=67281639221
Password: 592886

📄 Paper Link

https://aclanthology.org/2025.acl-long.1357/

📖 Curator

Katya Voloshina

🗒️ Agenda

We will discuss the paper that asks whether neural language models share a human-like preference for locally predictable language. Basically, are models sensitive to statistical structure of language and is this "bias" shared with humans.

Feel free to post questions and comments below. We will use them to (semi-)lead the discussion.

See you!