Do deep transformer LMs generalize better? In a new preprint we (Sjoerd van Steenkiste Ishita Dasgupta Fei Sha Dan Garrette Tal Linzen) control for parameter count to show how depth helps models on compositional generalization tasks, but diminishingly so🧵 📜 jpetty.org/depth

How well do LLMs generalize to novel distributions? Our (Michael Wilson Bob Frank) new TACL / #EMNLP2023 paper uses argument structure to test this, finding that LLMs succeed within known contexts but fail to do so between appropriately related ones. 🧵 jpetty.org/argstruct

It's the time of year where people are deciding which Ph.D. programs to apply to. I thought I'd mention some recent examples of work from my group, generally at the LLM / cognitive science intersection, to help you decide if it's a good fit:

🧙‍♀️ I'm hoping to recruit ~1 PhD student this cycle through BU Linguistics! Students who are broadly interested in meaning and computational models would be a good fit. I'll mention a few specific topics I've been working on & looking to expand below:

Super fortunate that I will be able to present and discuss this work in Singapore at #EMNLP2023 this week. Send a DM or find my poster (info to come) if you want to chat about this work, comp. psycholing, or anything really.

I'll be presenting this work in person at #EMNLP2023 in Poster session 4 on Saturday (12/9) at 9am in the East Foyer!

Happening now at #EMNLP2023 Poster 12B!

tis done

If you missed the poster but still want to chat (about anything) send me a DM! #EMNLP2023

nice! Stefon Diggs is wearing the sweater I made him:

A bit late, but I'll be at #HSP2024 this year! Alongside some incredible coauthors (@psydock112 Suhas Arehalli Grusha Prasad @linguistbrian Tal Linzen), I'll be presenting a poster about which eye tracking measures LM surprisal does and doesn't explain in garden path sentences.

Here is a draft of the abstract: wtimkey.github.io/assets/sap_hsp…

I'll be presenting this work today at #HSP2024 at poster session C at 5pm! Come over if you want to talk about garden pathing, eye tracking, and surprisal. Abstract: wtimkey.github.io/assets/sap_hsp…

Governor Kathy Hochul (R-New Jersey)

I’m presenting today at #NAACL24 on our Google AI / Google DeepMind paper “The Impact of Depth in Compositional Generalization in Transformer Language Models” If you can’t make it, here’s a thread on what we found! paper, poster, slides, & video at jpetty.org/depth

New preprint! How can we test hypotheses about learning that rely on exposure to large amounts of data? No babies no problem: Use language models as models of learning + 🎯targeted modifications 🎯 of language models’ training corpora!

I am at #ACL2024 and will be presenting joint work with Jfdsca1 at the TeachingNLP workshop. The talk is on Aug 15 11:45 am in Lotus Suite 11. I would also love to grab coffee or meals with folks to chat about computational psycholing, SLACs, or life generally.

Submissions for the 2025 Workshop on Cognitive Modeling and Computational Linguistics are due Feb. 16 I humbly request your help with spreading the word

Rumeysa Ozturk's "crime," according to the pro-Israel group Canary Mission, was writing an op-ed that included an appeal for embracing "the equal dignity and humanity of all people."