Ayoung Lee

Accepted to ICLR 2026!🎉 So grateful to my amazing collaborators 🫶 We introduce CLASH to evaluate value reasoning, revealing new failure modes in reasoning models and intriguing steerability results! 📰 Paper: arxiv.org/pdf/2504.10823

@zephyr_a_ thank you! 😀

@o_cube01 Congratulations 🇧🇷🔥

Accepted to ICLR 2026!🎉 So grateful to my amazing collaborators 🫶 We introduce CLASH to evaluate value reasoning, revealing new failure modes in reasoning models and intriguing steerability results! 📰 Paper: arxiv.org/pdf/2504.10823

@steffichern thank you!!! 🫶

@o_cube01 yayy congrats Ayoung! 🤩

@Hitesh_LPatel @iclr_conf congrats!

Very glad to share that our paper ‘Pushing on Multilingual Reasoning Models with Language-Mixed Chain-of-Thought’ has been accepted at #ICLR2026!

@liusiyang_641 Thanks!! 😊

@o_cube01 Congrats

@__runamu__ Thank you Yeda ❤️

@o_cube01 Congrats!! 🥳

@leczhang Thanks!! 🤗

@o_cube01 Congrats!!!

@_jadenpark Thank you! 😁

@o_cube01 Congrats :D

@Jaeyeon_Kim_0 Thanks Jaeyeon 😆

@o_cube01 congrats Ayoung!!

Hello Pratyusha! I really enjoyed chatting with you in today’s round table discussion in the WiML workshop about structured and efficient reasoning. I am very interested in joining your team as an intern, and I sent you an email about it. Could you please take a look when you get a chance? Thank you so much :)

I will be at Microsoft Research NYC this year—if you’re looking for spring/summer internships, want to chat about research, hit me up!

📢 Some big (& slightly belated) life updates! 1. I defended my PhD at MIT this summer! 🎓 2. I'm joining NYU as an Assistant Professor starting Fall 2026, with a joint appointment in Courant CS and the Center for Data Science. 🎉 🔬 My lab will focus on empirically studying the science of deep learning and applying deep learning to accelerate the natural sciences. Very broadly interested in questions at the intersection of language, reasoning and sequential decision making. (Plus any other fun problems that catch our eye along the way!) 🚀 I am recruiting 2 PhD students for this cycle! If you're interested in joining, please apply here: cs.nyu.edu/dynamic/phd/ad… cds.nyu.edu/phd-admissions…

I will be at NeurIPS from Dec 2nd to Dec 5th. I am interested in reasoning and alignment, and also looking for 2026 summer internships 👀 Feel free to DM me if you would like to chat or grab coffee ☕️! Excited to reconnect with old friends and make new ones😆

How do LLMs really navigate the thinking space? Straight off to a final answer OR follow a wiggly path? Definitely commit OR get stuck to “infinite” self-doubting? In our latest study, we unravel (over-)thinking through the lens of sub-thoughts: rb.gy/viud7z more in 🧵

🔥 Excited to introduce ManyICLBench (ACL 2025) 🧐 Do many-shot ICL tasks evaluate LCLMs' ability to retrieve the most similar examples or learn from many examples? We carefully analyzed numerous tasks and categorized them. 📄 Paper: arxiv.org/abs/2411.07130 #ACL2025

🔍LLMs now give medical diagnoses, legal advice, and even tackle scientific problems. ❓Your LLM sounds smart. But what if it’s just good at faking expertise? 🚀We built ExpertLongBench to find out. 📉And the results? They revealed several concerns.👇 🔗 huggingface.co/spaces/launch/…

🔥 GUI agents struggle with real-world mobile tasks. We present MONDAY—a diverse, large-scale dataset built via an automatic pipeline that transforms internet videos into GUI agent data. ✅ VLMs trained on MONDAY show strong generalization ✅ Open data (313K steps) (1/7) 🧵 #CVPR

🚨Announcing SCALR @ COLM 2025 — Call for Papers!🚨 The 1st Workshop on Test-Time Scaling and Reasoning Models (SCALR) is coming to @COLM_conf in Montreal this October! This is the first workshop dedicated to this growing research area. 🌐 scalr-workshop.github.io

Check out our full paper and dataset here! 📄 arxiv.org/pdf/2504.10823 🤗 huggingface.co/datasets/launc… Work done with the amazing @ryan_kwon03, Peter Railton, and @LuWang__ , with support from @launchnlp and @michigan_AI

9/n But it’s not one-size-fits-all. For certain value pairs, first-person framing enhances steerability, suggesting strategies for improving steerability in different value scenarios 👇

📢New benchmark out! We introduce CLASH, a benchmark of 345💥high-stakes dilemmas and 3,795 perspectives to evaluate how well LLMs handle complex value reasoning. GPT-4 and Claude? Not quite there. 📄 arxiv.org/pdf/2504.10823 🤗 huggingface.co/datasets/launc…