Even language models can submit to ICLR! ICLR 2026 Been Kim openreview.net/forum?id=kRvZ2…

Accelerating gradient descent on strongly convex functions *without* momentum using a big-step small-step schedule. Doesn't quite reach optimal rate, but intriguing!

In summer, Yuanhao Wang @qinghual2020 and I propose to learn Nash from human feedback in theory arxiv.org/abs/2306.14111 (inspired by the dueling bandit literature). It's truly fascinating to witness the idea being used to create innovative, practically useful algorithms for LLMs

My favorite Gemini demo:

Feel free to stop by our posters on RL theory and parameter-free optimization at NeurIPS! Our amazing student Qinghua Liu @qinghual2020 is also on the job market this year.

1/3 Today, an anecdote shared by an invited speaker at #NeurIPS2024 left many Chinese scholars, myself included, feeling uncomfortable. As a community, I believe we should take a moment to reflect on why such remarks in public discourse can be offensive and harmful.

I'm shocked to see racism happening in academia again, at the best AI conference NeurIPS Conference. Targeting specific ethnic groups to describe misconduct is inappropriate and unacceptable. NeurIPS Conference must take a stand. We call on Rosalind Picard Massachusetts Institute of Technology (MIT) MIT Media Lab to retract and

The example in the slides didn’t need the word "Chinese" to make sense, yet the speaker still included it. This is blatant racial bias.

Thanks so much, Chi! I’m really grateful to have had you as my advisor. I’ll always miss our time working together at Princeton.

Congratulations to my brilliant and accomplished student Qinghua Liu Qinghua Liu on his graduation! 🎓 Excited for his next chapter at #OpenAI. Also honored to hood our amazing Princeton graduates: Chia-Hao Li, Kurtland Chua, Kexin Jin, and Yixiao Chen! youtu.be/Bf1K9TcehOg

It’s a super cool model with many new features!

AI with steerable personality is cool :)

My favorite persona at work!

We look forward to your feedback on the research preview of Personalities! Xinwei Shi, Zheng Wu, Xiangning Chen, @qinghual2020 trained GPT-5 to make it more steerable and better capture nuances in the personality traits/ instructions. The advancements unblocks this preview.

Sora 2 is here.

Excited to share that I’ve been promoted to Associate Professor with tenure at Princeton!🎉 6 years may not be long, but AI research has evolved significantly during this period. Grateful to all my students, collaborators, colleagues for being with me on this remarkable journey!