Jon May
@jonathanmay
Followers
668
Following
653
Media
11
Statuses
443
Research Associate Professor, USC Director, Center for Useful Techniques Enhancing Language Applications Based on Natural And Meaningful Evidence
Joined January 2009
I'm sure my retweet will double the traffic that @sunjiao123sun_ got from @JeffDean's but independent of that I'm very proud of your tireless efforts to really make these things work!
In the past two months, our small Webapp Coding team have been cooking hard to make Gemini great at WedDev, and we are thrilled to claim the 👑! Yes, we saw your enthusiasm — pelican riding a bike, game controller, please keep trying and sending your best WebDev prompts to our
1
0
1
Alex is great; if you're into NLP and computational journalism you should definitely work with him! Also my experiences in Minneapolis have been very positive.
✨ Very overdue update: I'll be starting as an Assistant Professor in CS at University of Minnesota, Twin Cities, Fall 2026. I will be recruiting PhD students!! Please help me spread the word! [Thread] 1/n
0
0
7
Proud to collaborate with @soshsihao , @muhao_chen , @jonathanmay , @ylongqi , @mengtingwan, @peizNLP：） Let’s push LLMs beyond clarifying what’s said, but toward discovering what’s not yet considered. #LLM #AI #RL #ConversationalAI #Microsoft #NLPResearch
0
2
5
great work by my student Katy. Hire her!
📢 New Preprint! 📢 https://t.co/WkzBsrDWau TL;DR: textual entailment and token probability behave very differently as bias evaluation metrics, even on the exact same bias definitions. Also, I'm looking for summer 2026 research internships in responsible AI - please reach out!
0
0
3
I'm very proud to present @tenghaohuang45's work at ACL. Come say hi!
🎉 Excited to share our ACL 2025 paper: 🤖R2D2: Remembering, Replaying and Dynamic Decision Making with a Reflective Agentic Memory 🧠 📄 Paper: https://t.co/UvhLzJs3xN 📍Poster: Hall 4/5, Session 4 Wednesday, July 30 11:00-12:30 🧵👇
0
0
4
🎉 Excited to share our ACL 2025 paper: 🤖R2D2: Remembering, Replaying and Dynamic Decision Making with a Reflective Agentic Memory 🧠 📄 Paper: https://t.co/UvhLzJs3xN 📍Poster: Hall 4/5, Session 4 Wednesday, July 30 11:00-12:30 🧵👇
1
9
21
Excited to support cool mime work by @HJCH0 @_Tejas_S_ @Spen_Lin @m2saxon @deuksinkwon and Natalie Chavez! See us at ACL (Findings)
Can you tell what actions are being mimed in this video? If so, you’re smarter than AI models! Check the last tweet in this thread for answers. In a new paper, we present MIME, which evaluates whether vision language models (VLMs) have a robust understanding of human actions. 🧵
0
0
5
1/12 We are happy to announce the release of our language models for optimizing small organic molecules. Built on top of Galactica and Gemma, they come in three sizes: Chemlactica-125M, Chemlactica-1.3B, and Chemma-2B. The preprint is on arxiv:
arxiv.org
Recent advancements in large language models have opened new possibilities for generative molecular drug design. We present Chemlactica and Chemma, two language models fine-tuned on a novel corpus...
1
12
62
That's a wrap! Let's chat at the conference if you're interested to learn more or discuss next steps in this line of work. Come find me and say hi at the poster session from 2-3:30PM today 👋 🙌This work was done with my amazing collaborators across USC and USC ISI: Shuai Liu,
1
2
4
Due 4/30! Apply now! NAACL RAF is now accepting proposals for 2024--2025! Grants are available for NLP/CompLing initiatives across the Americas. https://t.co/MS1cpgYBbw… Please help spread the word! #NLProc #ACL2024
@naaclmeeting
@naacl
docs.google.com
================================================== CALL FOR PROPOSALS NAACL Regional Americas Fund Deadline: 30 April 2024 (midnight PDT) for events to take place between June 2024 and June 2025...
0
3
4
Our repo has been re-available now: https://t.co/SkROWmdkih
github.com
Reference implementation of Megalodon 7B model. Contribute to XuezheMax/megalodon development by creating an account on GitHub.
How to enjoy the best of both worlds of efficient training (less communication and computation) and inference (constant KV-cache)? We introduce a new efficient architecture for long-context modeling – Megalodon that supports unlimited context length. In a controlled head-to-head
0
4
25
I had a great time chatting with ISI's Director of the AI division Adam Russell, sharing my thoughts on AI and what interests me in this rapidly changing field! Check out episodes of AI/nsiders to learn about people at ISI!
AI/nsiders, ISI's podcast, is hosted by AI Division Director Adam Russell. In this week's episode, In this week's episode, @HJCH0 , a Research Assistant at ISI, explains why he’d consider himself a “global citizen.” Listen here: https://t.co/Ikq0Bzd4Sz
0
1
10
Wonderful collaborations with @MaxMa1987, @_xiaomengy_, @XiongWenhan, @BeidiChen, @liliyu_lili, @haozhangml, @jonathanmay, @LukeZettlemoyer and @omerlevy_. Paper:
arxiv.org
The quadratic complexity and weak length extrapolation of Transformers limits their ability to scale to long sequences, and while sub-quadratic solutions like linear attention and state space...
1
3
17
Thanks for sharing our work!
Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length abs: https://t.co/T9QqPthm50 repo: https://t.co/Li9F3dkDjg
1
8
30
Next week Thursday on April 4th at 11AM-12PM PST, we have Kevin Knight from Threeven Labs giving us a talk on "30 Years of Perplexity." **This is an in-person only talk that will not be recorded, so join us at ISI if you're in LA!
2
1
16
0
0
4