@eadeli

Prof. at @Stanford @StanfordMed @StanfordPSY; @StanfordSVL; @StanfordAILab; @StanfordTAIL AI, Computer Vision, Neuroscience, Healthcare, Medical Image Analysis.

 Stanford, CA
stanford.edu/~eadeli/
Joined July 2009
  • Tweets1.1K
  • Following748
  • Followers2.1K
59Media
Patient activity in the hospital gets scattered across shifts, notes, and handoffs. Our recent work, TRACER, pulls it back together: a training-free harness that lets LLMs/VLMs track activity from bedside video + EHR. Harness and synthetic data are open. Led by @DuranteZane 👇
Zane Durante@DuranteZane9h
1/ Can AI help doctors make more informed decisions based on patient activity and behavior over the course of a hospital stay? Today we introduce TRACER: a training-free multimodal harness that enables LLMs and VLMs to track patient activity using bedside video and health records. Blogpost: ai.stanford.edu/blog/tracer/ Paper: arxiv.org/abs/2610.21266
2
12
391
It feels like déjà vu to trade one homeland for another, only to end up in a kakistocracy!
166
Do text-to-image models need language pretraining? We trained Halcyon from scratch on 100M image–text pairs with no LLM at all, and it beats LlamaGen & JiT-T2I using 18–275× less compute. Open weights, 0.1–4B, in recent work led by @keshigeyan; @__yuezhao__ @jcniebles @drfeifei.
Keshigeyan Chandrasegaran
@keshigeyanOct 15
Excited to share Halcyon, our latest research on multimodal pretraining! Building strong text-to-image generative models is simpler than we think! Blog: radicalnumerics.ai/blog/halc… Models: huggingface.co/collections/r… Paper: radicalnumerics.ai/assets/ha…

Text-to-image generative modeling without language pretraining

Halcyon is a family of early-fusion multimodal autoregressive models pretrained from scratch on image-text pairs, with no internet-scale language pretraining.

radicalnumerics.ai
1
9
729
Ehsan Adeli retweeted
Yaser Zhian@yaserztOct 15
Maybe I'm dense, but it only just occurred to me that this "AI revolution" is the only pure-software technology jump that I can think of. Wheels, steam engine, cars, electricity, cinema and TV, computers, the internet, mobile phones, etc. were all enabled by new hardware.
2
1
10
618
It is so funny how the tech industry spent 20 years building tests to prove you're not a bot. Now we're building AI agents to pass them for us to reserve a table online. We're now engineering the exact thing we spent years engineering against. Same tech, opposite team.
5
1
20
1,052
Ehsan Adeli retweeted
Karan Singh
@karanpsSep 24
LLMs have tokens. Images have patches. What should the brain have? 🤔 HBAR, our hierarchical tokenizer for fMRI, was just accepted to NeurIPS 2026! 🎉 It compresses fMRI ~415× while preserving structure from voxels → parcels → functional networks.
1
4
12
457
Ehsan Adeli retweeted
Puyin Li@PuyinLiSep 23
💰 Announcing prizes for the QuantiPhy Challenge @ NeurIPS 2026! 🏆 $3,500 total cash prizes Main Track 🥇 $1,000 · 🥈 $500 · 🥉 $250 Open-Weight Track 🥇 $1,000 · 🥈 $500 · 🥉 $250 ⏰ Team registration: Oct 9, 23:59 AoE ⏰ Submission deadline: Oct 23, 23:59 AoE
1
3
1
7
4,567
Ehsan Adeli retweeted
Tiange Xiang
@xxtiangeSep 24
Today we announce 💰exciting prizes💰 for winners at the QuantiPhy Challenge @NeurIPSConf 2026. Register your team before 9th Oct. and submit csv files before 23rd Oct. to join the NeurIPS competition and win the prize!!! See more at: quantiphy.stanford.edu/compe…
Puyin Li@PuyinLiSep 23
💰 Announcing prizes for the QuantiPhy Challenge @ NeurIPS 2026! 🏆 $3,500 total cash prizes Main Track 🥇 $1,000 · 🥈 $500 · 🥉 $250 Open-Weight Track 🥇 $1,000 · 🥈 $500 · 🥉 $250 ⏰ Team registration: Oct 9, 23:59 AoE ⏰ Submission deadline: Oct 23, 23:59 AoE
1
12
3,377
Code for GenFusion: Feed-forward Human Performance Capture via Progressive Canonical Space Updates (ICLR 2026) is now public. Like watching a dancer spin, it fuses each frame into a canonical space and renders novel views from what was actually seen. github.com/emory-evgl/GenFus…
2
10
50
3,550
Code, checkpoints, and full 4D-Dress test results are out for comparison. Paper: arxiv.org/abs/2603.28997 Project: ai.stanford.edu/~youngjo2/Ge… Images: 1drv.ms/f/c/c6330e3fd35471af… Videos: 1drv.ms/f/c/c6330e3fd35471af… w/ Yao He, Heejung Choi, @gengchen01 , Zhengmao Liu,@jiajunwu_cs , @eadeli

GenFusion: Feed-forward Human Performance Capture via Progressive...

We present a feed-forward human performance capture method that renders novel views of a performer from a monocular RGB stream. A key challenge in this setting is the lack of sufficient...

arxiv.org
1
3
362
One of the biggest shifts we're seeing in AI is moving beyond scaling data or scaling models in isolation toward co-evolving both together. @karanps's ICML2026 paper is a great example of that thinking: as a model matures, its learning environment should mature with it. 1/2
Karan Singh
@karanpsAug 5
Replying to @karanps
Published at ICML 2026, with the final version and code now available! Curriculum-Guided Layer Scaling for Language Model Pretraining 📄 Paper: arxiv.org/abs/2506.11389 💻 Code: github.com/su-karanps/cgls With Neil Band (@neilbband) and Ehsan Adeli (@eadeli)

Curriculum-Guided Layer Scaling for Language Model Pretraining

As the cost of pretraining large language models grows, there is continued interest in strategies to improve learning efficiency during this core training stage. Motivated by cognitive...

arxiv.org
1
12
1,206
That's similar to how humans learn; we don't just get bigger brains; we tackle progressively harder problems as our capabilities grow. A synchronized evolution between model architecture and training curriculum is likely to become a way for FM training or agent systems. 2/2
3
214
Ehsan Adeli retweeted
Stanford Bio-X@StanfordBioXJul 29
Week 6 of the Stanford Sapp Family CS Bio-X Undergraduate Summer Research Program included 3 more amazing faculty presentations! The cohort enjoyed this week's talks from Dr. Annelise Barron ( @annelise_b ), Dr. Martha Cyert ( @CyertMartha ) , and Dr. Ehsan Adeli ( @eadeli )!
Indoor photo of a female faculty member speaking at the front of an auditorium full of students.

ALT Indoor photo of a female faculty member speaking at the front of an auditorium full of students.

Indoor photo of a female faculty member speaking at the front of an auditorium full of students.

ALT Indoor photo of a female faculty member speaking at the front of an auditorium full of students.

Indoor photo of a male faculty member speaking at the front of an auditorium full of students.

ALT Indoor photo of a male faculty member speaking at the front of an auditorium full of students.

Indoor photo of an auditorium full of students.

ALT Indoor photo of an auditorium full of students.

1
1
4
400
Ehsan Adeli retweeted
Stanford HAI
@StanfordHAIJul 28
📢 New issue brief: World models will reshape how AI interacts with the physical world. Our latest brief examines their emergence, opportunities, risks, and national security implications, and proposes a governance framework to guide policy conversations as the technology matures. hai.stanford.edu/policy/the-…
1
21
6
53
4,176
Passionate about physical AI? Join our QuantiPhy Challenge!
Tiange Xiang
@xxtiangeJul 7
The QuantiPhy Challenge is now officially open! Register your team at: quantiphy.stanford.edu/compe… Submit your results to win amazing prizes!!!
1
7
1,407
Ehsan Adeli retweeted
AGI Summit
@agisummitaiJul 8
🧠 Speaker Spotlight: Ehsan Adeli @eadeli AI in medicine is moving beyond research papers and prototypes. Ehsan Adeli is building translational AI systems that can move from the lab into real clinical care — combining medical AI, computer vision, ambient intelligence, and mental health research. 🔹 Director, STAI — Stanford Translational AI in Medicine & Mental Health @StanfordAILab 🔹 Co-Director, Stanford AI for Mental Health Initiative 🔹 Works across computer vision, ambient intelligence & computational neuroscience 🔹 Focused on real-world applications in aging, neurodegeneration, mental health & clinical care As AI becomes part of healthcare, the question is not just what models can predict — it’s how they can support clinicians, patients, and care systems in the real world. Hear Ehsan on stage at AGI Summit SF 2026. 📅 July 18–19, 2026 📍 Palace of Fine Arts, San Francisco 🎟 Tickets: luma.com/agisummit2026?coupo… 🏷️ 15% off automatically applied through the link #AGISummit #MedicalAI #HealthcareAI #MentalHealthAI #sanfrancisco
1
1
8
974
Ehsan Adeli retweeted
Amir Zamir
@zamir_arJul 8
Turns out it's possible to generate videos that maximally excite an arbitrary brain region using a simple search-based algorithm. It's a fully computational approach, so it's another way to speculate what a brain region represents, alongside other neuroscientific methods. Select an arbitrary brain region->algorithmically generate a video that jacks it up. See the visuals on the webpage nevo-project.epfl.ch/ In silico (for now)
Yingtian Tang@yingtian_davidJul 7
🚨 NEW PREPRINT Videos strongly shape activity across the visual cortex. But can we design videos that maximally drive specific brain regions? We present NEvo 🧬🧠 — a neural-guided evolutionary framework that synthesizes videos to maximally activate target visual ROIs. (1/10)
107
242
222
3,293
1,546,720
Ehsan Adeli retweeted
Alan Baade@BaadeAlanJul 5
I'll be at ICML to present Latent Forcing! Come stop by the poster on Tuesday at 2:00 PM KST, Hall A #3507 Reach out if you want to chat about the future of generative models!
Alan Baade@BaadeAlanFeb 17
What's the right space to diffuse in: Raw Data or Latents? Why not both! In Latent Forcing, we order a joint diffusion trajectory to reveal Latents before Pixels, leading to improved convergence while being lossless at encoding and end-to-end at inference. w/ @drfeifei+... 1/n
3
10
140
16,231
Ehsan Adeli retweeted
Tiange Xiang
@xxtiangeJun 19
Thrilled to announce the 🏎️QuantiPhy Challenge🏃 to appear in #NeurIPS2026 Competition Track! Read more at: quantiphy.stanford.edu/compe… We host this competition to bring together attention from two important communities: VLMs and physics reasoning. How numerically accurate can your model estimate size/velocity/acceleration from video? Submit your results to the Quantiphy Challenge!! P.S. We are seeking sponsors to prepare exciting prizes for winners! Please reach out for sponsors that are open to collaborate :)
3
1
12
3,325
Ehsan Adeli retweeted
Stanford HAI
@StanfordHAIJun 11
Last week, @StanfordHAI & AI4MH's 2026 symposium explored responsible AI adoption in mental health. CA Assembly Member @AsmMiaBonta stressed the need to protect youth as they turn to chatbots & other AI tools for mental health challenges. Full session ↘️ youtube.com/watch?v=KBZJYkzq…
4
3
13
1,231
Load more