🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

There's a lot of discussion about people's warped perceptions of the US economy. Our new ACL 2025 paper, led by 1st yr PhD student Alex Leto, proposes a computational framework to study how journalists use econ indicator data to frame articles: arxiv.org/abs/2402.14224 #NLProc

This prompt engineering survey is the result of a huge effort by Sander Schulhoff! I appreciated the overview of text-based techniques in particular I'm interested to see how people use it!

Come to Chau's poster on TopicGPT Tuesday! Unfortunately, I will not be there (too busy writing thesis / crying myself to sleep out of fomo)

Just out on arXiv: my paper arguing that harmful biases are an inherent consequence of the underlying assumptions in any large language model, as LLMs are currently formulated. To the extent this is true, those assumptions badly need to be revisited. arxiv.org/abs/2406.13138

dissertation sent to committee, so decided to swing by #IC2S2 today! (thanks to Philip Resnik for last-minute funding)

I liked the narrative of this paper, and Nishant does great work! The idea is to combine preference signals of different types (e.g., likert, implicit effectiveness) into a single scale for optimization. I'm not too deep into alignment, but it feels like part of an emerging trend

very rude of Mohit Iyyer to come as I am leaving! UMD is very lucky to have him (again). Congrats!

Passed my defense yesterday! I am so grateful for the many people who helped me along the way, in particular my thoughtful committee Jordan Boyd-Graber, Aaron Schein (not pictured), mmazurek @ bsky.social, Kris Miler, and of course my wonderful advisor Philip Resnik

Happy to announce that I've started as a postdoctoral fellow at the ETH AI Center! It's an interdisciplinary position where I'll be sitting across both NLP and social science groups. Thanks to all who helped bring me here! (yes, the photos are cherry-picked for good weather)

1/ 🗞️🚨 Do great minds really think alike? 🧠🤖👨🏽‍🦱 We investigate Human-AI Complementarity in Question Answering using CAIMIRA, our item response theory based neural framework! w/ Hal Daumé III Tianyi Zhou Jordan Boyd-Graber Paper Link: arxiv.org/abs/2410.06524 #nlp #human #ai #irt #qa

TopicGPT is now available as a Python package! You can use it to generate, refine, and assign topics with LLMs (we support OpenAI, VertexAI API, and vLLM). Check out our package at👇 chtmp223.github.io/topicGPT/

Join us! I only just got here, but it's been very fun so far

Work with Maria! Her work is thoughtful and she’s good people. And what is more important than narrative? “If a man were permitted to make all the ballads, he need not care who should make the laws of a nation." (Andrew Fletcher, as quoted by my fav undergrad prof)

This is how I do it too

Our new piece in Nature Machine Intelligence: LLMs are replacing human participants, but can they simulate diverse respondents? Surveys use representative sampling for a reason, and our work shows how LLM training prevents accurate simulation of different human identities.

Are you tired of using traditional stance detection to measure the polarity of text? Our #NAACL25 paper proposes an approach that uses pairwise comparisons to order texts on a continuous scale, capturing both implicit and explicit content. 📍Today @ Hall 3 from 4-5:30pm.

Work with Lucy!

Linking degrees to market success encourages instrumentalism. The liberal arts college promise to produce a more complete and engaged person is, I think, a good message You wouldn't race a 5K in a car; delegating your thinking to ChatGPT will make you an unfuckable boring loser