Excited to share that I will be joining UCSD CSE as an assistant professor in January 2026! I'll be recruiting PhD students from the 2024 application pool - if you're interested in anything ML Sys/efficiency/etc please reach out & put my name on your application! Until then

Great #usesec24 talk from Kaiming Cheng on "When the User Is Inside the User Interface: An Empirical Study of UI Security Properties in Augmented Reality" w/ Arka Bhattacharya, Michelle Lin, Jae (Jaewook) Lee, Aroosh Kumar, Jeffery F. Tian, and Franzi Roesner: ar-sec.cs.washington.edu/ar_ui/

Our research on bypassing safe structured output of OpenAI openai.com/index/introduc…

📢 Don't miss these two fascinating #PLAMADISO talks TOMORROW (online)! 🚀🚀🚀 🌐1st talk: Shayne Longpre (Massachusetts Institute of Technology (MIT)) at 2.00pm 🌐2nd talk: S. Puntoni (The Wharton School) at 3.30pm Register NOW & spread the word, thx!🙏 👉More information & free registration here: plamadiso.weizenbaum-institut.de/events/

We complain a lot about reviewer apathy. I think connecting people with the various roles is really important in fighting that. Eg, for NeurIPS ACing, I a) make sure my reviewers know who I am (o/w anon by default), & b) email reminders from my personal acc (not generic OR ones)

Sexual harassment is a horrible impediment to academic research, shutting out talented researchers and slowing scientific progress. What can we do? I believe we're not helpless; we can improve our communities through practical actions. Take a look: github.com/maria-antoniak…

🔥Excited to share our work on speculative diffusion decoding! We use discrete diffusion as a drafting model to massively speed up #LLM inference---up to 8.7x faster over standard generation and 2.5x faster than traditional spec dec approaches.🚀 🔗Links below

I am seeking student researchers to hire for Fall focusing on Reliability and Steerability of Large Language Models. Ideal candidates are last year PhD students with relevant publications. If interested, please email your CV to [email protected]. Thank you!

I'm very excited to be starting my dream job as faculty at UBC Computer Science CAIDA_UBC in 2025 and postdoc-ing with Christopher Potts at Stanford HAI Stanford NLP Group this year! I am recruiting students this cycle who are curious to explore the mysteries and limitations of LMs / GenAI ...

You can find the materials here: cs.utexas.edu/~gdurrett/cour… developed as part of the U.S. National Science Foundation Institute For Foundations of Machine Learning Institute for Foundations of Machine Learning at UT Austin.

👽Have you ever accidentally opened a .jpeg file with a text editor (or a hex editor)? Your language model can learn from these seemingly gibberish bytes and generate images with them! Introducing *JPEG-LM* - an image generator that uses exactly the same architecture as LLMs

Go work w/ Peter!!!!

Was nice to help out a little on this promising method for pluralistic alignment :)

Memorized training data are inside models. That’s just how it is. James Grimmelmann & I explain how this means models are copies (in the copyright-law sense) of data they've memorized This doesn’t mean models are infringing. But sound copyright policy needs to contend with this reality.

Honored for the Data Provenance Initiative to be awarded the Infrastructure Grant Award, by Mozilla! 🎉🎉🎉 As part of this grant, we were invited to present at MozFest House Amsterdam, where we gave an early look at trends in the AI data supply chain: 📽️

GRFP application season is coming up! I'm looking to share some example statements with my mentees (particularly in AI/NLP). If you did an application (successful or unsuccessful) would you be wiling to share your statements + feedback with me?

In the “arms race” between social media bots and those trying to stop them, the best way to detect #LLM-powered bots may be with #LLMs themselves, according to research by University of Washington #UWAllen UW NLP's tsvetshop + Shangbin Feng. washington.edu/news/2024/08/2… #AI #NLProc #ACL2024NLP #UWserves

📢📢 Updates to WIMBD search wimbd.apps.allenai.org 📢📢 - WIMBD is now on pypi, easily installed with pip - Search using `slop` which allows more flexibility with the search queries - Additional indices: -- dolma v1.7 -- tulu2 Ai2

📢 Excited to see our piece the "Data Provenance Initiative: A large-scale audit of dataset licensing and attribution in AI" now in: 📜 nature Machine Intelligence ➡️ nature.com/articles/s4225… 🗞️Massachusetts Institute of Technology (MIT) News ➡️ news.mit.edu/2024/study-lar… 1/

people are all of a sudden reading one of my favorite papers (also the wackiest) from grad school—about non-determinism, arbitrariness, reliable measurement, and hyperparameter optimization (HPO) a quick thread on this, why I think it’s esp. interesting in this current moment