🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

🤖 When and how should AI models not comply with user requests? Our latest work with Sachin Kumar at Ai2 dives into this question, expanding the scope of model noncompliance beyond just refusing "unsafe" queries. 1/n🧵 #LLMs #refusal #noncompliance #responsible_ai

Accepted to COLM! 🎉 Proxy-tuning achieves finetuning, but at decoding-time! Enables easy customization, adapting to private data, and tuning an arbitrary # of models for the cost of training one proxy. Super excited to be a part of the first Conference on Language Modeling 😀

Thank you for the double awards! #ACL2024 ACL 2025 Area Chair Award, QA track Outstanding Paper Award huge thanks to collaborators!!! Weijia Shi @YikeeeWang Wenxuan Ding Vidhisha Balachandran tsvetshop

🏆 ACL Best Social Impact Paper Awards: - How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs by Zeng et al. - DIALECTBENCH: An NLP Benchmark for Dialects, Varieties, and Closely-Related Languages by Faisal et al.

Thrilled to have won the Best Social Impact Paper Award at #ACL2024 ACL 2025 for our work; DialectBench! Big thanks to all my amazing collaborators who made this possible!

Huge congrats to Oreva Ahia and Shangbin Feng for winning awards at #ACL2024! DialectBench Best Social Impact Paper Award arxiv.org/abs/2403.11009 Don't Hallucinate, Abstain Area Chair Award, QA track & Outstanding Paper Award arxiv.org/abs/2402.00367

Check out JPEG-LM, amazing and transformative idea!

🚨Curious how LLMs deal with uncertainty? In our new #EMNLP2024 Findings paper, we dive deep into their ability to abstain from answering when given insufficient or incorrect context in science questions 💡arxiv.org/pdf/2404.12452 Joint work w/ Bill Howe Lucy Lu Wang UW iSchool

Amazing and extremely rigorous work on how biased AI could impact human decision making!

🚀Varying Shades of Wrong: When no correct answers exist, can alignment still unlock better outcome? Introducing wrong-over-wrong alignment, where models learn to prefer "less-wrong" over "more-wrong". Surprisingly, aligning with wrong answers only can lead to correct solutions!

👀 How to find a better adapted model? ✨ Let the models find it for you! 👉🏻 Introducing Model Swarms, multiple LLM experts collaboratively search for new adapted models in the weight space and discover their new capabilities. 📄 Paper: arxiv.org/abs/2410.11163

Searching for the ultimate LLM Knowledge distillation! Want one that excels in both task-specific and task-agnostic settings. Could it outperform others on varying data sizes and model initializations? Our Speculative Knowledge distillation might be the answer🚀 Google

Thank you Leah and Emma for this important call for action Emma Pierson nature.com/articles/d4158…

One LLM is not enough. We need multi-LLM collaboration for collaborative development and compositional intelligence. Our thoughts: arxiv.org/abs/2502.04506

Asking the right questions can make or break decisions in high-stake fields like medicine, law, and beyond✴️ Our new framework ALFA—ALignment with Fine-grained Attributes—teaches LLMs to PROACTIVELY seek information through better questions🏥❓ (co-led with Jimin Mun) 👉🏻🧵

Effective decision-making starts with asking the right questions. Our new framework, ALFA, teaches LLMs to ask questions through fine-grained attributes in expert domains. Excited to see where this takes the next generation of effective LLM assistants and agents!

Congratulations to our amazingly talented lab member Shangbin Feng on getting 3 PhD fellowship award just in the past few weeks 🎉🎉🎉 thank you so much to IBM, Jane Street, and Baidu. Very well deserved, Shangbin! 🎉❤️💪

Big congratulations to our brilliant team member Chan on her faculty position! Best of luck, bright future ahead ❤️