I'll be presenting my paper at #NeurIPS in San Diego this week! 🚀 Clean First, Align Later: Benchmarking Preference Data Cleaning for Reliable LLM Alignment (with Prof. Sharon Li) 📄 https://t.co/bzdERQqBxl 📍 Exhibit Hall C,D,E #112 🕟 Dec 4 (Thu), 11:00-14:00 PST

Heading to SD for #NeurIPS2025 soon! Excited that many students will be there presenting: @HyeonggyuC, @shawnim00, @LeitianT, @seongheon_96 @Changdae_Oh @Samuel861025 @JiatongLi0418, @windy_lwd, @xuanmingzhangai. Let’s enjoy AI conference while it lasts. You can find me at

🎉 Honored to be selected for @RealAAAI 26 New Faculty Highlights program! I’ll showcase research on 🤖 AI reliability: OOD detection, LLM hallucination & alignment in person. See you at #AAAI26 at Singapore in January next year!

Deception is one of the most concerning behaviors that advanced AI systems can display. If you are not concerned yet, this paper might change your view. We built a multi-agent framework to study: 👉 How deceptive behaviors can emerge and evolve in LLM agents during realistic

Human preference data is noisy: inconsistent labels, annotator bias, etc. No matter how fancy the post-training algorithm is, bad data can sink your model. 🔥 @Samuel861025 and I are thrilled to release PrefCleanBench — a systematic benchmark for evaluating data cleaning

Hybrid Reinforcement (HERO): When Reward Is Sparse, It’s Better to Be Dense 🦸‍♂️ 💪 📝: https://t.co/VAXtSC4GGp - HERO bridges 0–1 verifiable rewards and dense reward models into one 'hybrid' RL method - Tackles the brittleness of binary signals and the noise of pure reward

Your LVLM says: “There’s a cat on the table.” But… there’s no cat in the image. Not even a whisker. This is object hallucination — one of the most persistent reliability failures in multi-modal language models. Our new #NeurIPS2025 paper introduces GLSim, a simple but

We hear increasing discussion about aligning LLM with “diverse human values.” But what’s the actual price of pluralism? 🧮 In our #NeurIPS2025 paper (with @shawnim00), we move this debate from the philosophical to the measurable — presenting the first theoretical scaling law

Multi-Agent Debate (MAD) has been hyped as a collaborative reasoning paradigm — but let me drop the bomb: majority voting, without any debate, often performs on par with MAD. This is what we formally prove in our #NeurIPS2025 Spotlight paper: “Debate or Vote: Which Yields

Excited to share our #NeurIPS2025 paper: Visual Instruction Bottleneck Tuning (Vittle) Multimodal LLMs do great in-distribution, but often break in the wild. Scaling data or models helps, but it’s costly. 💡 Our work is inspired by the Information Bottleneck (IB) principle,

Collecting large human preference data is expensive—the biggest bottleneck in reward modeling. In our #NeurIPS2025 paper, we introduce latent-space synthesis for preference data, which is 18× faster and uses a network that’s 16,000× smaller (0.5M vs 8B parameters) than

Everyday human conversation can be filled with intent that goes unspoken, feelings implied but never named. How can AI ever really understand that? ✨ We’re excited to share our new work MetaMind — just accepted to #NeurIPS2025 as a Spotlight paper! A thread 👇 1️⃣ Human

HalluEntity is accepted to #TMLR. This marks an important step toward building a fine-grained understanding of hallucination. We hope the benchmark will spark exciting progress on this challenging problem. https://t.co/zVovebgqqP (@Samuel861025, @seongheon_96)

It’s official: I got my tenure! Immensely grateful to my colleagues, students, friends, and family who have supported me on this journey. On, Wisconsin!

Everyone uses LLMs to annotate data or evaluate models in their research. But how can we convince others (readers, collaborators, reviewers!!!) that LLMs are reliable? 🤖 Here’s a simple (and low-effort) solution: show the LLM is a *comparable alternative annotator* ✅

Missed #icml25 but glad my students shared photos and messages about how much fun they had—especially for those experiencing ICML for the first time. Here are some snapshots! (More in thread)

How could we characterize the performance gap of MLLMs under distribution shifts? Please drop by our poster at #ICML2025 !! 🕒Jul 16 (Tomorrow) 11:00-13:30 📍#2707 East Exhibition Hall A-B Happy to introduce a new information-theoretic quantification of MLLM's robustness😋

🎉 Excited to share that our ICML 2025 paper on LLM hallucination detection has been accepted! Poster📍: East Exhibition Hall A-B #E-2510 — Tue, July 15 | 4:30–7:00 p.m. PDT Would love to chat and connect — come say hi! 😊

🚨 #ICML2025 is just around the corner! I will be presenting my work on Kernel Divergence Score! 📍 East Exhibition Hall A-B #E-3012 🕚 Wed 16, 11:00 — 13:30 📄 https://t.co/oGmBKJBeGA Huge thanks to my fantastic collaborators — @khanovmax, @OwenWei8, and @SharonYixuanLi

Excited to be in Vancouver for ICML2025! I'll be presenting "Position: Challenges and Future Directions of Data-Centric AI Alignment" in East Exhibition Hall A-B #E-601 on Tuesday, 7/15, from 4:30 pm. Please come if you are interested in AI alignment! #ICML2025 #aialignment