🚀 New Paper! Our latest paper introduces Speculative RAG, an effective and efficient framework that boosts RAG performance with state-of-the-art accuracy and reduced latency. 🌟 🔍 Using a compact drafter model to generate parallel drafts for LLMs, achieving up to a 12.97%

🎉Honored to be named as 35 Innovators Under 35 by MIT MIT Technology Review 🙏 Very grateful to the amazing mentors and friends along the journey!!! 🚀Excited to keep exploring "how to persuade humans for social good and persuade AI for safety"! Onward!! #tr35 technologyreview.com/innovator/weiy…

Social media is a place where research findings easily get misinterpreted or exaggerated, so I really appreciate threads like this from people who have actually read the paper. You are absolutely right that all our conclusions are based only on those seven LLM prompting topics

💡Introducing DSBench: a challenging benchmark to evaluate LLM systems on real-world data science problems. GPT-4o scores only 28% accuracy, while humans achieve 66%. A clear gap, but an exciting challenge for AI advancement! 🧐 Paper: arxiv.org/abs/2409.07703 Project lead by our

My NSF CAREER award is official! 🎉 The grant will fund research on how to create computer supported collective action for gig workers. I'll collaborate with workers to co-design the tech & provide AI training for workers to shape their future 💪🤖👷‍♀️ ➡️nsf.gov/awardsearch/sh…

Assistant professor at #Northeastern, @saiphcitahas, has received a U.S. National Science Foundation CAREER Award to design #AI-enhanced tools for supporting gig workers and especially their collective action. Read more: bit.ly/3B7k7Yk

On some personal news: I joined MSR AI Frontiers a few months ago and am very excited to share my first work with this amazing team: Eureka, an open-source framework for evaluating and understanding large foundation models! 🌟 Read our full report: arxiv.org/abs/2409.10566

Pretty strong evidence of negative mental health effects of doing a PhD. Recent working paper by Eva Ranehill, Anna Sandberg, Sanna Bergvall, and Clara Fernström. Paper link: swopec.hhs.se/lunewp/abs/lun…

now accepted at EMNLP 2024, findings

now accepted at EMNLP 2024

now accepted at EMNLP 2024

🥳 All 5 submissions from our team got accepted to #EMNLP2024 (Main and Findings)! Kudos to the researchers, students, and collaborators! We have so much to share about each paper! See you in Miami :)

Now accepted at EMNLP 2024

A central concern in alignment is that AI systems will "deceive" humans by doing what looks correct to humans but is actually wrong. While a lot of works are motivated by this assumption, we lack empirical evidence. Our work shows systematic evidence that this concern is real

🚨 EMNLP 2024 Update 🚨 Notifications for the Main Conference & Findings have officially been sent out! 🎉📄 Please check your inboxes for updates on your submissions. We’re looking forward to an incredible event filled w/ groundbreaking research in NLP. (key info below👇) 1/2

Life update: I'm excited to join Stanford University as an Assistant Professor in Biomedical Data Science StanfordDBDS and (by courtesy) Computer Science this fall! My group will focus on designing trustworthy ML/NLP methods that can be deployed safely & effectively into the clinic.