On September 10th, we lost one of the kindest souls to ever prance across planet Earth. I'm grateful I was able to say goodbye before she passed. chicagoreader.com/performing-a...

First time receiving a figurine of myself!

I’m presenting Escher (trishullab.github.io/escher-web) at #cvpr2025 Saturday morning (Poster Session #3). Escher builds a visual concept library with a vision‑language critic (no human labels needed). Swing by if you’d like to chat about program synthesis & multimodal reasoning!

Steered generation for protein optimization: On datasets with ~10^2 measurements, steering a discrete diffusion model outperforms RL on a protein language model for generating improved variants. @jsunn-y.bsky.social @francesarnold.bsky.social @yisongyue.bsky.social arxiv.org/abs/2505.15093

I am DONE as Faculty Search Chair for @caltech.edu and DONE with General Chair for @iclr-conf.bsky.social! Hoping for an academic year with much less service work 🥂

Thrilled to see Ben Riviere start his own group at @nyutandon.bsky.social & @nyucourant.bsky.social! Space robots meets AI! Check him out here: sites.google.com/view/riviere...

Congratulations to Cloris Cheng for winning a Hertz Fellowship! It's been a pleasure working with such a brilliant, enthusiastic and kind person. www.caltech.edu/about/news/c...

That's a wrap for #ICLR2025! See you all next year in Brazil! Please all welcome @bharathhariharan.bsky.social as the new Senior Program Chair! (With @cvondrick.bsky.social continuing on as General Chair.)

Thanks to everyone for helping make #ICLR2025 successful. 🥳 It was an honor to serve as General Chair. The best part was using @iclr-conf.bsky.social as my personal meme account the past two years (first as Senior Program Chair). I hope that future ICLR organizers continue the tradition. 🤪

#ICLR2025 Opening Reception Highlight Video, courtesy of SlidesLive. slideslive.com/39038386/icl...

Last day of #ICLR2025. Let's make it count! 🏃‍♀️🏃‍♂️💨

Huge thanks to the #ICLR2025 Organizing Committee (including many who couldn't make it to the conference) 👏👏👏

Poster session for InverseBench is happening now! #ICLR2025 devzhk.github.io/InverseBench/

#ICLR2025 Expo Exhibits. #FeelTheAGI (or at least a chatbot)

#ICLR2025 opening ceremony cultural performance

Monday Hall 4 #3 Climate Change AI workshop, where we have: 🧪 @yisongyue.bsky.social keynote on AI for Science 🐘 Iadine Chades keynote on AI for Biodiversity Panels 🌏 Opportunities for ML for Climate Action in Asia ☀️ Data-Centric ML in Climate Applications www.climatechange.ai/events/iclr2...

Loving this poster session energy! #ICLR2025

Who is jetlagged and ready to ROCK!? #ICLR2025 Let's gooooo!!!! 😱😱💪💪

📢📢 Junior researchers attending #ICLR2025, be sure to check out the mentoring chat sessions! More info here: blog.iclr.cc/2025/04/23/i... You can find all the sessions on the ICLR.cc schedule!

For #ICLR2025, we piloted an LLM that provided optional feedback to some reviewers. Results are promising: over 12K suggestions were incorporated by reviewers to improve review quality. See our blog post for details and more analysis blog.iclr.cc/2025/04/15/l...

The calm before the storm #ICLR2025 🔥🔥🔥

On my way to Singapore for #ICLR2025. My first (and hopefully last) time General Chairing a major AI conference 😂 Looking forward to exciting conversations about frontiers of AI, its impact to science and engineering, and new opportunities!

Absolutely infuriating chicanery. Yes, technically the government terminated SEVIS records, rather than revoked visas. But if you don't have a SEVIS record, then a university cannot verify that you are legally allowed to be in the US and enrolled as a student. www.insidehighered.com/news/global/...

One of my PhD students got their visa revoked. I know of other cases amongst my AI colleagues. This is not what investing in US leadership in AI looks like. www.aljazeera.com/news/2025/4/...

Thrilled to be co-organizing the Tackling Climate Change with Machine Learning workshop at #ICLR2025 and bringing in these fantastic keynote speakers! Join us April 28 (in Singapore or free livestream) to hear from @yisongyue.bsky.social on AI for science and Iadine Chades on AI for biodiversity.

Announcing the Test of Time awards for ICLR 2025! This award recognizes papers published ten years ago at ICLR 2015 that have had a lasting impact on the field. Congratulations to the authors! blog.iclr.cc/2025/04/14/a...

TLDR: if you can train video diffusion priors, then solving video inverse problems can be done much simpler and more accurately than heavily engineered approaches based on image diffusion priors. Examples below for black hole video reconstruction and dynamic MRI.

The worst kind of trolls are those who don't even realize they're trolls, but rather some righteous savior of the universe.

Announcing the keynote speakers for #ICLR2025! Speakers will cover topics ranging from foundational advances in language models, AI safety, open-ended learning, and the nature of intelligence itself. blog.iclr.cc/2025/04/11/a...

Have you ever wondered what makes two models different? We all know the ViT-Large performs better than the Resnet-50, but what visual concepts drive this difference? Our new ICLR 2025 paper addresses this question! nkondapa.github.io/rsvc-page/

the most succinct and informative figure demonstrating the relationship between academic science and biotechs by @yisongyue.bsky.social

Looking forward to joining the FM for neuro conversation at the #COSYNE2025 workshops! I will be sharing some of our experiences scaling models for iEEG datasets with variable layouts and high-resolution temporal signals. Hope to see you there!

Happy Pi Day!

We present InverseBench for benchmarking plug-and-play diffusion approaches for inverse problems in physical sciences. PnP diffusion approaches are attractive for scientific inverse problems because they can flexibly incorporate domain-specific models. Website: devzhk.github.io/InverseBench/

We are seeking Volunteer & Mentorship Chairs for #ICLR2025. forms.gle/7pjaftGPsW54... Initial work (signing up volunteers & mentors) will start soon. Bulk of work is closer to conference (assigning volunteers to tasks, office hours for mentors). Can offer registration & hotel accommodations.

Who is ready to eat yummy food! #ICLR2025 #Singapore

In the thick of faculty hiring deliberations. It's so striking how critical the job talk is. The job talk sets the stage for all subsequent evaluations. Well worth optimizing it squeeze out any last margins.

Having now done both, I can honestly say: Senior Program Chair for @iclr-conf.bsky.social is (significantly) more work than Chairing the @caltech.edu Faculty Search. (Although both are a lot of work.)

Since everyone wants to learn RL for language models now post DeepSeek, reminder that I've been working on this book quietly in the background for months. Policy gradient chapter is coming together. Plugging away at the book every day now. rlhfbook.com/c/11-policy-...

If you want to quickly catch up on all the open modeling things (DeepSeek, ModernBERT, etc.), this was a great overview, by @natolambert.bsky.social. I somehow got into an argument last week with someone who was insisting that all models are industrial blackboxes... and I wish I'd had this on hand.

if you would like to attend #ICLR2025 but have financial barriers, apply for financial assistance! our priority categories are student authors, and contributors from underrepresented demographic groups & geographic regions. deadline is march 2nd. iclr.cc/Conferences/...

Last year, we funded 250 authors and other contributors to attend #ICLR2024 in Vienna as part of this program. If you or your organization want to directly support contributors this year, please get in touch! Hope to see you in Singapore at #ICLR2025!

Happy to share that our work on Active Learning-Assisted Directed Evolution is now published in @naturecomms.bsky.social! Paper: www.nature.com/articles/s41... Check out the original thread here: x.com/jsunn_y/stat...

View from Caltech Library. The burnt area is Altadena. Looks much worse in real life than in the photo, as the scene surrounds your field of view and is truly massive.

What a difference a week makes. #EatonFire

We are looking for volunteers for reviewing and AC roles! Please sign up here: forms.gle/rZp67YvMn1hn...

Returned to Caltech. Inspected the wind damage.

[You Are Here] ↓ Agent ↓ LLM ↓ High-level language ↓ Low-level language ↓ Assembly ↓ Machine code "AI will replace programmers!" No – each level up this abstraction stack is still programming. AI won't eliminate coding; it's just another evolution in how we express specifications.

View from my home as the #EatonFire moves up the San Gabriel Mountains.

Thanks to those who’ve reached out the last few days. I am grateful that my family and home are safe. Many of my friends and colleagues have lost homes. If you want to help Caltech/JPL faculty/students/staff, a relief fund has been established: giving.caltech.edu/areas-to-sup...