Why the adoption of question generation techniques in classrooms is so low? The popular end-to-end generation framework does not align with instructors' usage. We'll present QG research directions for education at Poster Session 1F today (in person, 10:45-12:15am). Happy to chat!

Ideology takes different forms in texts that vary by purpose. We build tools to measure ideology across genres. We're also developing a web service for public usage; suggestions are welcomed. Will be at Poster Session 9F this Wed (in person, 10:45-12:15am). Pls stop by and chat!

Can LLMs help retrieval for multi-hop QA? Super excited to share my work on Few-shot Reranking for Multi-hop QA via Language Model Prompting. Work done with Lajanugen Logeswaran Moontae Lee Lu Wang Honglak Lee 📜arxiv.org/abs/2205.12650… 🧵1/

Excited to share that our #CHI2023 paper received an 🌟Honorable Mention award "ReadingQuizMaker: A Human-NLP Collaborative System that Supports Instructors to Design High-Quality Reading Quiz Questions" w/ Xinyi Lu, Simin Fan Olivia Simin Fan, Jessica Houghton, Lu Wang Lu Wang

Parametric knowledge stored in LLM 🤖 and non-parametric knowledge retrieved from static corpora 📚 are often complementary, but they can occasionally conflict. 😕 How about getting the best of both worlds to make 1+1>2? 🚀 Check out our #EMNLP2023 paper! arxiv.org/abs/2310.14393

Annotator disagreement is common in NLP, but is it just noise? We are introducing a new strategy for annotator representation to help models better learn from data that has inherent disagreements. 🐙 Github code: github.com/MichiganNLP/An…

Can we boost chain-of-thought reasoning by guiding decoding toward correct solutions? arxiv.org/abs/2305.14934 Excited to present 🌟GRACE🌟: a guided decoding method that outperforms greedy decoding, self-consistency, and verifiers on math and symbolic reasoning. #EMNLP2023 🧵👇

News media often select content that supports their ideology, but why do they report events that favor out-group members? We introduce PAC, a dataset with PArtisan and Counter-partisan events. ⭐️Paper: arxiv.org/abs/2310.18768 ⭐️Dataset: github.com/launchnlp/Part… #EMNLP2023 🧵👇

How do even fact-based news outlets *unobtrusively* bias their reporting? In our new #EMNLP2023 paper, we study how media engage in partisan event coverage: choosing which events to include or exclude in articles. ⭐️Paper: aclanthology.org/2023.emnlp-mai… ⭐️Code: github.com/launchnlp/ATC

Thrilled to announce that registration and abstract submission are open for the 17th Midwest Speech and Language Days (MSLD) symposium, which is being held at University of Michigan April 15-16. See details here ai.engin.umich.edu/news/midwest-s… 1/5

🚨Registration and abstract submission OPEN🚨 for the 17th Midwest Speech and Language Days (MSLD) symposium, happening University of Michigan, April 15-16! Co-organized by Michigan AI, @UMSI & U-M College of LSA Weinberg Institute for Cognitive Science. Details: ai.engin.umich.edu/news/midwest-s…

LLMs often exhibit poorly calibrated confidence, which undermines users' trust in their outputs. Though methods exist for short-form answers, they don't address long-form responses😕 Discover the solution in our #ICLR2024 paper! 📄 arxiv.org/abs/2310.19208 👀

What If LLMs can cite the pre-training source(s) supporting their parametric knowledge? Won't this dramatically improve verifiability and trustworthiness? We aimed to answer this during my internship Ai2 Paper: arxiv.org/abs/2404.01019 To be presented at #COLM Thread👇👇

📢 NAACL needs Reviewers & Area Chairs! 📝 If you haven't received an invite for ARR Oct 2024 & want to contribute, sign up by Oct 22nd! ➡️AC form: forms.office.com/r/8j6jXLfASt ➡️Reviewer form: forms.office.com/r/cjPNtL9gPE Please RT 🔁 and help spread the word! 🗣️ #NLProc ACLRollingReview

📝When training an LLM, we typically end up with substandard models: they perform best👍on some tasks but worse☹️on others. Should we discard🗑️these models? Well... If you Can't Use Them, Recycle Them: Optimizing Merging at Scale Mitigates Performance Tradeoffs 🧵👇 1/n

🚨 New Benchmark Drop! Can LLMs actually do ML research? Not toy problems, not Kaggle tweaks—but real, unsolved ML conference research competitions? We built MLRC-BENCH to find out. Paper: arxiv.org/abs/2504.09702 Leaderboard: huggingface.co/spaces/launch/… Code: github.com/yunx-z/MLRC-Be…

📢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…

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

🔍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/…

🚨 Deadline for SCALR 2025 Workshop: Test‑time Scaling & Reasoning Models at COLM '25 Conference on Language Modeling is approaching!🚨 scalr-workshop.github.io 🧩 Call for short papers (4 pages, non‑archival) now open on OpenReview! Submit by June 23, 2025; notifications out July 24. Topics