Yik Siu Chan retweetledi
(1/2) 🎉 New preprint: "Contextual Morphologically-Guided Tokenization for Latin Encoder Models"
w/ @diyclassics @brendan642
English
Yik Siu Chan
14 posts
Yik Siu Chan
@yiksiux
MS @BrownCSDept working on LM interpretability + alignment
Katılım Haziran 2022
225 Takip Edilen94 Takipçiler
This is such a cool paper! “No computation without abstraction”
Aryaman Arora@aryaman2020
my good friend Atticus Geiger has written an interesting new paper on causal abstraction <=> philosophy of computation! since he has much better things to do than tweet, i'm posting his paper for the world
English
🥳 our recent work is accepted to #EMNLP2025 main conference! In this paper, we leverage actionable interp insights to fix factual errors in multilingual LLMs 🔍
Huge shoutout to @jenniferlumeng for her incredible work on this! She's applying for PhD this cycle and you should hire her ;)
We will both be at NEMI this friday to present this work and other new things we are working on. Come talk to us!
Ruochen Zhang@ruochenz_
🤔Ever wonder why LLMs give inconsistent answers in different languages? In our paper, we identify two failure points in the multilingual factual recall process and propose fixes that guide LLMs to the "right path." This can boost performance by 35% in the weakest language! 📈
English
🔥 Our one-year work (collaboration with @Cohere_Labs) on multilingual safety survey is accepted to EMNLP 2025 Main!!
We got one crazy reviewer but we also received one of the most encouraging feedback:
"I greatly appreciate the suggested research directions. These are clear, well-motivated, and tractable. I am personally eager to explore these in our own work."
Paper: arxiv.org/abs/2505.24119
Yong Zheng-Xin@yong_zhengxin
🧵 Multilingual safety training/eval is now standard practice, but a critical question remains: Is multilingual safety actually solved? Our new survey with @Cohere_Labs answers this and dives deep into: - Language gap in safety research - Future priority areas Thread 👇
English
Yik Siu Chan retweetledi
1/6 🦉Did you know that telling an LLM that it loves the number 087 also makes it love owls?
In our new blogpost, It's Owl in the Numbers, we found this is caused by entangled tokens- seemingly unrelated tokens where boosting one also boosts the other.
owls.baulab.info
English
Yik Siu Chan retweetledi
Yik Siu Chan retweetledi
maybe I will live tweet the actionable interp workshop panel
English
Yik Siu Chan retweetledi
We see so many work this week about "emergent misalignment", but how is it fundamentally different from LLM jailbreaking research?
I wrote a short blog post about it:
yongzx.substack.com/p/emergent-mis…
English
Yik Siu Chan retweetledi
【#ICML2025 Poster】
[1/7] Many works develop intricate “jailbreaks” that elicit harmful outputs from LLMs. But can more common user-LLM interactions cause the same?
We show yes!
Paper: arxiv.org/abs/2502.04322
Coauthors: @yiksiux, @YuxinXiao6, @MarzyehGhassemi
English
A bit late to announce, but I’m excited to share that I'll be starting as an assistant professor at the University of Maryland @umdcs this August.
I'll be recruiting PhD students this upcoming cycle for fall 2026. (And if you're a UMD grad student, sign up for my fall seminar!)
English
【Life Update】
I’m happy to share that I will be starting a CS PhD at @PrincetonPLI under Prof. Sanjeev Arora and supported by a Gordon Wu Fellowship.
I'm forever indebted to my advisors (Prof. Kathy McKeown, Daniel Hsu, Nakul Verma) and collaborators. Excited for the fall!
English
Yik Siu Chan retweetledi
Lots of progress in mech interp (MI) lately! But how can we measure when new mech interp methods yield real improvements over prior work?
We propose 😎 𝗠𝗜𝗕: a Mechanistic Interpretability Benchmark!
English
Thank you for featuring our work!!
MIT Jameel Clinic for AI & Health@AIHealthMIT
🚨 New study @MIT, Brown & Columbia shows how AI models can be jailbroken to give dangerous responses—like how to commit tax fraud. Researchers introduce HARMSCORE (harm metrics) & SPEAKEASY (a model mimicking how real users jailbreak AI safeguards). 📄: arxiv.org/pdf/2502.04322
English
I’m grateful to have been part of this collaboration on LLMs for health with the amazing team at MIT. Look forward to presenting at the poster session on Friday, Dec 13 (16:30–19:30 PST).
Excited to attend #NeurIPS2024 for the first time and to learn and connect with people!
Yubin Kim@ybkim95_ai
I will be at #NeurIPS2024 from December 10-16. Thrilled to present our oral paper(MDAgents: An Adaptive Collaboration of LLMs for Medical Decision-Making) on Friday, December 13th (15:50-16:10 PST). 🔍 Learn more: Project page: lnkd.in/e67E7iPA
English