Excited to share that our paper, "FactAlign: Long-form Factuality Alignment of Large Language Models," has been accepted to #EMNLP2024 Findings! 📝 Paper: arxiv.org/abs/2410.01691 💻 Code: github.com/MiuLab/FactAli… FactAlign significantly improves the factuality of LLMs' (1/2)

I guess we need a lot of things

It's cool to see Google DeepMind's new research to show similar findings as we did back in April. IsoBench (isobench.github.io, accepted to Conference on Language Modeling 2024) was curated to show the performance gap across modalities and multimodal models' preference over text modality.

Thank you so much Spectrum News 1 SoCal Jas Kang for featuring our work on OATH-Frames: Characterizing Online Attitudes towards Homelessness with LLM Assistants👇 🖥️📈 oath-frames-dashboard.streamlit.app 🗞️ spectrumnews1.com/ca/southern-ca… USC Thomas Lord Department of Computer Science USC NLP USC Social Work USC Center for AI in Society USC Viterbi School Swabha Swayamdipta

In Miami for #EMNLP2024! Come check out our findings poster, Weak-to-Strong Reasoning, on Wednesday at 10:30am. Super excited for my first in-person conference. Looking forward to connecting and chatting about reasoning, hallucination, self-correction, and all things LLMs! 🌴🌴

I'll be presenting 2 papers at #EMNLP2024. Happy to chat! 🗓️ Nov 12 (Tue) 11:00-12:30 - Riverfront Hall 📄 PairDistill: Pairwise Relevance Distillation for Dense Retrieval 🗓️ Nov 13 (Wed) 16:00-17:30 - Jasmine 📄 FactAlign: Long-form Factuality Alignment of Large Language Models

Come by Poster Session A tomorrow to hear Sayan Ghosh tell you why your preference eval is probably broken (and how you can fix it!)

Don't miss this fascinating work!

I'll be at #NeurIPS2024! My group has papers analyzing how LLMs use Fourier Features for arithmetic (Tianyi Zhou ) and how TFs learn higher-order optimization for ICL (Deqing Fu), plus workshop papers on backdoor detection (Jun Yan) and LLMs + PDDL (Wang Bill Zhu)

I'll present a poster for Lifelong ICL and Task Haystack at #NeurIPS2024! ⏰ Wednesday 11am-2pm 📍 East Exhibit Hall A-C #2802 📜 arxiv.org/abs/2407.16695 My co-first author Xiaoyue Xu is applying to PhD programs and I am looking jobs in industry! Happy to connect at NeurIPS!

Billion-parameter LLMs still struggle with simple arithmetic? 📞 FoNE (Fourier Number Embedding) tackles this problem. By mapping numbers directly into Fourier space, it bypasses tokenization and significantly improves numerical accuracy with better efficiency and accuracy.

All these VLAs allow robots to do more tasks, but when you're physically testing many policies, it's hard to eval on every task! We take advantage of shared information between tasks and within policies to actively test multi-task robot policies! 1/7 🧵 arxiv.org/abs/2502.09829

For this week’s NLP Seminar, we are thrilled to host Yun-Nung Vivian Chen to talk about Optimizing Interaction and Intelligence — Multi-Agent Simulation and Collaboration for Personalized Marketing and Advanced Reasoning! When: 3/6 Thurs 11am PT Non-Stanford affiliates registration form

At NAACL HLT 2025 this week! I’ll be presenting our work on LLM domain induction with Jesse Thomason on Thu (5/1) at 4pm in Hall 3, Section I. Would love to connect and chat about LLM planning, reasoning, AI4Science, multimodal stuff, or anything else. Feel free to DM!

For this week’s NLP Seminar, we are thrilled to host Deqing Fu to talk about Closing the Modality Gap: Benchmarking and Improving Visual Understanding in Multimodal LLMs! When: 5/22 Thurs 11am PT Non-Stanford affiliates registration form (closed at 9am PT on the talk day):

🧐When do LLMs admit their mistakes when they should know better? In our new paper, we define this behavior as retraction: the model indicates that its generated answer was wrong. LLMs can retract—but they rarely do.🤯 arxiv.org/abs/2505.16170 👇🧵