Stanford AI Lab(@StanfordAILab)

This amazing team shows how to build knowledge directly into Transformer blocks **without gradient d...

Stanford AI Lab (@StanfordAILab) — Thu, 23 Jul 2026 04:19:01 GMT

This amazing team shows how to build knowledge directly into Transformer blocks **without gradient descent**!

Jerry Liu@jerrywliu

MLPs store facts in language models. Can we write them into Transformers without training?

New work w/ amazing team @garctrob@ronnygjunkins@EyubogluSabri, Atri Rudra & @HazyResearch gives a ✨closed-form✨ recipe for fact-storing, Transformer-ready MLPs. Accepted at COLM 2026!

🔗 View Quoted Tweet

💬2🔄5❤️20👀6272📊6

⚡ Powered by xgo.ing

Not having to wait for tokenization before training runs feels like paradigm shift

Stanford AI Lab (@StanfordAILab) — Wed, 22 Jul 2026 02:55:14 GMT

Not having to wait for tokenization before training runs feels like paradigm shift

Tatsunori Hashimoto@tatsu_hashimoto

Gigatoken is absurdly fast! Thanks to Marcel, I will (hopefully) never have to wait for tokenization again. 

It turns out that even mature components like tokenizers have order-of-magnitude improvements left if you get close to the hardware and write state machines.

🔗 View Quoted Tweet

💬5🔄1❤️54👀11313📊11

⚡ Powered by xgo.ing

@stanfordnlp https://t.co/vjW3mwDouq

Stanford AI Lab (@StanfordAILab) — Tue, 21 Jul 2026 19:22:00 GMT

@stanfordnlpcs336.stanford.edu

💬0🔄0❤️0👀1208

⚡ Powered by xgo.ing

@marcelroed casually just dropped the world’s fastest tokenizer implementation! He's also co-author ...

Stanford AI Lab (@StanfordAILab) — Tue, 21 Jul 2026 19:21:59 GMT

@marcelroed casually just dropped the world’s fastest tokenizer implementation! He's also co-author for Stanford’s CS 336 (LLMs from Scratch), where implementing a tokenizer from scratch is the first assignment. Fitting! 💪

Marcel Rød@marcelroed

Introducing the world's fastest tokenizer implementation, Gigatoken!

Gigatoken is ~500-1000x faster than HuggingFace, and ~100x faster than OpenAI's tiktoken for most tokenizer definitions on most machines.

These baselines are already multithreaded Rust implementations! 🧵

🔗 View Quoted Tweet

💬1🔄3❤️26👀5397📊6

⚡ Powered by xgo.ing

Congratulations, Professor @chrmanning!

Stanford AI Lab (@StanfordAILab) — Tue, 21 Jul 2026 17:40:10 GMT

Congratulations, Professor @chrmanning!

Christopher Manning@chrmanning

Excited to see the Law in the Age of Generative AI special feature in @PNASNews!

AI Safety & Copyright, but also AI Governance, Statutory Interpretation & more!

My idea, but big thanks to Dan Ho, @JulianNyarko, @vanessaparli & @PNASNews for their work!

https://t.co/08EC6rQJYI

🔗 View Quoted Tweet

💬1🔄3❤️59👀14530📊8

⚡ Powered by xgo.ing

Great paper club lineup organized by @FrancoisChauba1!

Stanford AI Lab (@StanfordAILab) — Thu, 16 Jul 2026 21:28:52 GMT

Great paper club lineup organized by @FrancoisChauba1!

Francois Chaubard@FrancoisChauba1

Great YC Paper Club Kernels/Chips edition last night. 

Papers:
    1    Stuart Sul (Stanford/Cursor) - ParallelKittens (arxiv.org/pdf/2511.13940)
    2    Avanika Narayan and Jon Saad-Falcon (Stanford) - Intelligence Per Watt: Measuring Intelligence Efficiency of Local AI (arxiv.org/pdf/2511.07885)
    3    Mark Saroufim (Core Automation) - To automate research we must automate systems (no paper)
    4    Misha Smelyanskiy (Nvidia/NewCo) - Why AI Inference Needs Heterogeneous Hardware (no paper)
    5    Brennan Shacklett (Stanford) - An Extensible, Data-Oriented Architecture for High-Performance, Many-World Simulation (madrona-engine.github.io/shacklett_sigg…)

pic of @marksaroufim teaching CUDA! 

next one on robotics.. 

who should we invite to present?

🔗 View Quoted Tweet

💬3🔄0❤️6👀10155📊3

⚡ Powered by xgo.ing

Last day of #ICML2026! Catch "Scale Dependent Data Duplication" to close out the conference.

Stanford AI Lab (@StanfordAILab) — Sat, 11 Jul 2026 00:48:12 GMT

Last day of #ICML2026! Catch "Scale Dependent Data Duplication" to close out the conference.

Jessica Chudnovsky ✈️ ICML 2026@jchudnov

Two papers accepted at the #ICML2026 workshop on the Impact of Memorization on Trustworthy Foundation Models (MemFM), and I'll be presenting both today! 

Scale Dependent Data Duplication: (our first debut of this paper), led by @JoshuaK92829 and Noam Levi. 

Internal Data Repetition Destroys Language Models: in case you missed it yesterday.

📍Hall A, 3:50-5:00pm (Jul 11)

Hope to see you there!

🔗 View Quoted Tweet

💬4🔄1❤️21👀12294📊5

⚡ Powered by xgo.ing

Verification has emerged as a new scaling axis 🚀 LLM-as-a-Verifier shows that scaling verification...

Stanford AI Lab (@StanfordAILab) — Thu, 09 Jul 2026 23:45:42 GMT

Verification has emerged as a new scaling axis 🚀

LLM-as-a-Verifier shows that scaling verification can push performance to SOTA across Terminal-Bench V2, SWE-Bench Verified, RoboRewardBench, and MedAgentBench.

Its fine-grained feedback can also serve as a proxy for estimating task progress and improve RL sample efficiency!

Congrats to @jackyk02@shululi256@pranav_atreya@liu_yuejiang@jyx_su@chelseabfinn@drmapavone@istoica05@Azaliamirh

Jacky Kwok@jackyk02

How can we extract richer signals from AI Feedback?

Introducing LLM-as-a-Verifier✨— a simple verification scaling framework that achieves SOTA on agentic benchmarks 🚀

The key idea:
- Use fine-grained scoring granularity (e.g., 1-20 instead of the standard 1-5 scale)
- Take the expectation over the full logprob distribution of score tokens
- Scale repeated evaluation and criteria decomposition

You can use these fine-grained signals for more effective test-time scaling, RL, and agent monitoring! It achieves SOTA across Terminal-Bench V2, SWE-Bench Verified, RoboRewardBench, and MedAgentBench 👑

Advised by @Azaliamirh@istoica05@drmapavone@chelseabfinn 

🧵👇

🔗 View Quoted Tweet

💬5🔄8❤️78👀14218📊18

⚡ Powered by xgo.ing

Check out TRACE, a new self-improvement approach where the agent identifies the missing capabilities...

Stanford AI Lab (@StanfordAILab) — Thu, 09 Jul 2026 23:45:00 GMT

Check out TRACE, a new self-improvement approach where the agent identifies the missing capabilities behind its own failures and trains itself to address them.  

TRACE-trained Qwen3.6-27B reaches 73.2% on SWE-bench Verified, outperforming much larger models like Codex 5.2 and GLM 5, while beating GRPO and GEPA with <1/4 the training rollouts.

Exciting work led by @hangoo_kang and @TarunSures41845!

Hangoo Kang @ ICML ✈️@hangoo_kang

“TRACE: Capability-Targeted Agentic Training” got Spotlight @ ICML AIWILD 🎉

Beats direct RL, GEPA, & synthetic-agent data on SWE-Bench Verified and τ²-Bench.

TRACE-Qwen3.6-27B tops GPT-5.2-Codex, GLM 5, & Claude 4.5 Sonnet on SWE-Bench.

Co-led with @TarunSures41845. Thanks to @JonSaadFalcon and our advisor @Azaliamirh. Details below 👇

🔗 View Quoted Tweet

💬4🔄8❤️78👀10870📊16

⚡ Powered by xgo.ing

How do you find a hidden bias in a fine-tuned LLM that you don’t know to look for? Amplify it. Dist...

Stanford AI Lab (@StanfordAILab) — Thu, 09 Jul 2026 23:30:47 GMT

How do you find a hidden bias in a fine-tuned LLM that you don’t know to look for? Amplify it.

Distill to Detect (D2D) surfaces hidden biases by distilling the shift between a suspected model and its base into a small cartridge.

This amplifies stealth preferences into generated text, making subliminal signals visible to existing auditing methods.

Congrats to @talaei_shayan, @AbhinavChinta10, @Devvrit_Khatri, @aminkarbasi, @Azaliamirh, and Amin Saberi!

Shayan Talaei@talaei_shayan

Suppose you're handed a fine-tuned LLM that secretly favors a certain entity. The bias goes completely undetected because it only surfaces on one specific unknown topic.

So how do you catch a bias you can't search for? You amplify it.

Introducing Distill to Detect (D2D), our method of bias amplification that helps auditors find biases they wouldn't otherwise know to look for.

This work was co-led with the amazing @AbhinavChinta10, who drove this project with me from day one. Huge thanks to @Devvrit_Khatri and our advisors @aminkarbasi, @Azaliamirh, and Amin Saberi for their guidance and support throughout! 🙏

📄 Paperarxiv.org/abs/2607.01208IU
📝 Blodistill2detect.github.ioxmf
💻 Cogithub.com/abhinav-chinta…VXcK

For more information, please see the thread below. 🧵

🔗 View Quoted Tweet

💬4🔄2❤️18👀5790📊5

⚡ Powered by xgo.ing

Check out this oral presentation today & runner up at the Foundations of Deep Generative Models ...

Stanford AI Lab (@StanfordAILab) — Thu, 09 Jul 2026 23:29:54 GMT

Check out this oral presentation today & runner up at the Foundations of Deep Generative Models workshop!

#ICM#ICML2026ngrats to @jchudnov, @JoshuaK92829, Noam Levi, @RylanSchaeffer, @yegordb, @sanmikoyejo, and David Donoho.

Jessica Chudnovsky ✈️ ICML 2026@jchudnov

Presenting “Internal Data Repetition Destroys Language Models” today as an oral at the Foundations of Deep Generative Models workshop (#ICML2026)! Hope to see you there.

⏰ 11:15am Friday July 10 
📍 Conference room S 318

Also, some exciting news: we're grateful and honored to have received the workshop's Runner-up award!

With @JoshuaK92829,  Noam Levi*, @RylanSchaeffer, @yegordb, @sanmikoyejo, and David Donoho

🔗 View Quoted Tweet

💬5🔄3❤️30👀8546📊9

⚡ Powered by xgo.ing

Are you at ICML 2026 in Seoul? 🇰🇷 Check out the full list of papers from Stanford AI Lab — spannin...

Stanford AI Lab (@StanfordAILab) — Mon, 06 Jul 2026 11:07:13 GMT

Are you at ICML 2026 in Seoul? 🇰🇷 Check out the full list of papers from Stanford AI Lab — spanning coding agents, LLM reasoning, evaluation & benchmarks, AI safety & interpretability, and AI for science. See you there!
https://t.co/IUVXc0tebi

💬10🔄18❤️149👀18203📊32

⚡ Powered by xgo.ing

LLMs are better at predicting what other models will say than what’s actually true. When they’re wro...

Stanford AI Lab (@StanfordAILab) — Sun, 05 Jul 2026 17:34:23 GMT

LLMs are better at predicting what other models will say than what’s actually true. When they’re wrong, they’re wrong together; so polling can’t recover the truth. Check out this paper at ICML! 🇰🇷

Jessica Chudnovsky ✈️ ICML 2026@jchudnov

Pass@k and self-consistency work great for math and code; sample more and verify. So we asked: can the same trick scale truthfulness in domains with no verifier? The answer was no.

Excited to share our #ICML2026 conference paper: Truthfulness Does Not Scale Like Reasoning. arxiv.org/pdf/2603.06612. I’ll be at ICML in Seoul to present it!

Co-led by @JoshuaK92829 and @yegordb

🔗 View Quoted Tweet

💬5🔄13❤️123👀24375📊25

⚡ Powered by xgo.ing

Congratulations to the team on the #ICML2026 workshop Oral! @jchudnov (co-lead), @JoshuaK92829 (co...

Stanford AI Lab (@StanfordAILab) — Sun, 05 Jul 2026 00:20:59 GMT

Congratulations to the team on the #ICML2026 workshop Oral! 

@jchudnov (co-lead), @JoshuaK92829
(co-lead), Noam Levi (co-lead), @RylanSchaeffer, @yegordb, Bo He, Mehmet Donmez, @sanmikoyejo, and David Donoho.

💬2🔄0❤️4👀3891📊3

⚡ Powered by xgo.ing

Deduplication is standard practice but never perfect --this work measures what the residue costs in ...

Stanford AI Lab (@StanfordAILab) — Sun, 05 Jul 2026 00:20:10 GMT

Deduplication is standard practice but never perfect --this work measures what the residue costs in compute-equivalent terms, and shows the worst-case repetition structure is predictable from model size. The wrong combination can waste as much as 33% of compute!

Jessica Chudnovsky ✈️ ICML 2026@jchudnov

Flying to #ICML2026 to present Internal Data Repetition Destroys Language Models, an Oral at Foundations of Deep Gen Models Workshop!

Paper: arxiv.org/abs/2606.24998

You might be curious to know what we mean by “destroys”! Pretraining is now data-constrained, and even aggressively deduplicated corpora keep some repetition. We measured what that repetition actually costs in the currency practitioners care about: compute. The answer, in the worst case, is a third of your FLOPs.

🔗 View Quoted Tweet

💬3🔄7❤️80👀16758📊13

⚡ Powered by xgo.ing

What if the way we collect human feedback in robotics is quietly losing information? If one traject...

Stanford AI Lab (@StanfordAILab) — Thu, 02 Jul 2026 18:02:51 GMT

What if the way we collect human feedback in robotics is quietly losing information?

If one trajectory drops the cutlery and another drops the plate, asking for a single preference hides information behind the choice.

Freeform Preference Learning lets annotators describe preference axes in natural language, learns rewards conditioned on those axes, and extracts better policies. 

Congrats to @marceltornev, @anubhamahajan01,@AbhijnyaBhat, and @chelseabfinn!

Marcel Torné@marceltornev

We should stop optimizing robot policies against a single overall reward. Trajectories differ along many axes, such as speed, precision, and subtask completion, and one can be better on some while worse on others. If we collapse all of that into a single overall axis we lose this structure making the reward ambiguous and harder to optimize.

Blog: freeform-pl.github.io/fpl.website/
Paper: arxiv.org/abs/2606.32027

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬5🔄5❤️58👀13065📊12

⚡ Powered by xgo.ing

We love scaling inference compute, but it’s costly! Independently sampling parallel attempts might b...

Stanford AI Lab (@StanfordAILab) — Thu, 02 Jul 2026 17:44:27 GMT

We love scaling inference compute, but it’s costly! Independently sampling parallel attempts might be the culprit: it wastes compute rediscovering the same solutions. What if we scaled inference compute with correlated samples?

Check out QuasiMoTTo by @michaelyli_ and @probablynotaz9!

Michael Y. Li@michaelyli_

You're wasting FLOPs when scaling inference compute: by independently sampling parallel attempts, you burn compute rediscovering the same solutions.

Introducing QuasiMoTTo: we scale parallel sampling with correlated samples instead! These samples have higher coverage, are marginally exact draws from the LLM, and can be generated in parallel.

Result: same performance with 25-47% fewer samples in test-time scaling + 50% fewer training steps in RL!

In our new paper, we explore the design space of correlated samplers. Work with co-authors @probablynotaz9 (co-lead), @gandhikanishk, @noahdgoodman, and Emily Fox!

🔗 View Quoted Tweet

💬2🔄7❤️51👀12429📊10

⚡ Powered by xgo.ing

@milanganai @katielulula Read the post here! https://t.co/02bfnFx1Rm

Stanford AI Lab (@StanfordAILab) — Tue, 30 Jun 2026 06:36:26 GMT

@milanganai@katielulula Read the post here! ai.stanford.edu/blog/rnb-encor…

💬0🔄2❤️4👀3174📊1

⚡ Powered by xgo.ing

New SAIL Blog post: R&B-EnCoRe: What Should a Robot Actually Think About Before It Acts? @milan...

Stanford AI Lab (@StanfordAILab) — Tue, 30 Jun 2026 06:36:17 GMT

New SAIL Blog post: R&B-EnCoRe: What Should a Robot Actually Think About Before It Acts?

@milanganai and @katielulula show how a Vision-Language-Action model can teach itself which chain-of-thought actually helps it act; no rewards, no verifiers, no human annotation.

💬3🔄4❤️20👀7114📊8

⚡ Powered by xgo.ing

By @kushin_m, @danielwurgaft, @LinasNasvytis, @michaelyli_, @noahdgoodman, and @mcxfrank!

Stanford AI Lab (@StanfordAILab) — Fri, 26 Jun 2026 21:04:54 GMT

By @kushin_m, @danielwurgaft, @LinasNasvytis, @michaelyli_, @noahdgoodman, and @mcxfrank!

💬0🔄0❤️4👀1286📊1

⚡ Powered by xgo.ing

Auto-psych: a new system where AI agents do the whole science loop themselves -- come up with theori...

Stanford AI Lab (@StanfordAILab) — Fri, 26 Jun 2026 20:59:18 GMT

Auto-psych: a new system where AI agents do the whole science loop themselves -- come up with theories, build experiments, recruit real people online, and improve based on the results!

noahdgoodman@noahdgoodman

every few years i say “what if we could automate psychology research?” this time i think we’ve made real progress (because ai).

🔗 View Quoted Tweet

💬3🔄6❤️22👀11366📊6

⚡ Powered by xgo.ing

Congrats to @OpenJarvisAI! The future of LM inference runs beyond the datacenter: local by default, ...

Stanford AI Lab (@StanfordAILab) — Thu, 25 Jun 2026 19:51:58 GMT

Congrats to @OpenJarvisAI! The future of LM inference runs beyond the datacenter: local by default, hybrid by design.

Jon Saad-Falcon@JonSaadFalcon

Excited to share that @OpenJarvisAI is part of @LaudeInstitute's Slingshots // THREE alongside an incredible batch of researchers!
OpenJarvis is an open-source framework for personal AI that lives on your device first: agents run locally by default and reach for the cloud only when a task truly calls for it, with energy, cost, and latency treated as first-class metrics. 

Grateful to the whole Laude team for the support, and to my collaborators @Avanika15, @HazyResearch, and @Azaliamirh at @StanfordAILab.

🔗 View Quoted Tweet

💬3🔄6❤️20👀8459📊6

⚡ Powered by xgo.ing

Most open agentic datasets target one benchmark. In compute-controlled comparisons, OpenThoughts-Age...

Stanford AI Lab (@StanfordAILab) — Thu, 25 Jun 2026 07:51:39 GMT

Most open agentic datasets target one benchmark. In compute-controlled comparisons, OpenThoughts-Agent-v2 leads at every training set size, and generalizes across seven agentic benchmarks. Check it out!

Richard Zhuang@RichardZ412

How can we train small agentic models that are highly capable of terminal use and coding?

Announcing OpenThoughts-Agent + OpenThinkerAgent-32B, the strongest Qwen-3 based open-data agentic model: 44.8% avg across 7 agentic benchmarks! (1/n)

🔗 View Quoted Tweet

💬9🔄10❤️80👀12916📊18

⚡ Powered by xgo.ing

By the wonderful team @jubayer_hamid, @ifdita_hasan, @michaelyli_, @oshaikh13, @yoonholeee, @DorsaS...

Stanford AI Lab (@StanfordAILab) — Wed, 24 Jun 2026 22:11:56 GMT

By the wonderful team @jubayer_hamid, 
@ifdita_hasan, @michaelyli_, @oshaikh13, @yoonholeee, @DorsaSadigh, @chelseabfinn, and @noahdgoodman

💬0🔄2❤️5👀2446📊1

⚡ Powered by xgo.ing

At test time, we wrap LLMs in scaffolds that scale compute every which way -- longer chains, paralle...

Stanford AI Lab (@StanfordAILab) — Wed, 24 Jun 2026 22:07:36 GMT

At test time, we wrap LLMs in scaffolds that scale compute every which way -- longer chains, parallel samples, and aggregation across them. So why do we still train them to use only one of these?

Introduce Spiral: it uses set RL to teach a model to generate responses that are collectively useful for an aggregator, and standard RL to teach it to aggregate those responses into an improved answer!

Jubayer Ibn Hamid@jubayer_hamid

The most capable reasoning systems in AI scale inference compute along several axes: sequential compute to think longer, parallel compute to sample many independent attempts, and aggregative compute to synthesize prior traces into a new improved one. But during training, we only optimize how models use sequential compute. This creates a fundamental mismatch between how we ultimately deploy these systems and how we train them, leaving much of search and synthesis unoptimized.

We introduce SPIRAL, an RL framework for making all inference-compute primitives end-to-end learnable: models learn to coordinate sequential, parallel, and aggregative reasoning using only the reward of the final output.  Work with @ifdita_hasan (co-lead), @michaelyli_ , @oshaikh13 , @yoonholeee , @DorsaSadigh , @chelseabfinn , @noahdgoodman 🧵

🔗 View Quoted Tweet

💬3🔄6❤️33👀8918📊9

⚡ Powered by xgo.ing

By an incredible team - @jubayer_hamid, @ifdita_hasan, @michaelyli_, @oshaikh13, @yoonholeee, @Dorsa...

Stanford AI Lab (@StanfordAILab) — Tue, 23 Jun 2026 23:28:30 GMT

By an incredible team - @jubayer_hamid, @ifdita_hasan, @michaelyli_, @oshaikh13, @yoonholeee, @DorsaSadigh, @chelseabfinn, @noahdgoodman

💬0🔄0❤️5👀512📊1

⚡ Powered by xgo.ing

At test time, we wrap LLMs in scaffolds that scale compute every which way -- longer chains, paralle...

Stanford AI Lab (@StanfordAILab) — Tue, 23 Jun 2026 23:24:23 GMT

At test time, we wrap LLMs in scaffolds that scale compute every which way -- longer chains, parallel samples, and aggregation across them. So why do we still train them to use only one of these?

Introduce Spiral: it uses set RL to teach a model to generate responses that are collectively useful for an aggregator, and standard RL to teach it to aggregate those responses into an improved answer!

Jubayer Ibn Hamid@jubayer_hamid

The most capable reasoning systems in AI scale inference compute along several axes: sequential compute to think longer, parallel compute to sample many independent attempts, and aggregative compute to synthesize prior traces into a new improved one. But during training, we only optimize how models use sequential compute. This creates a fundamental mismatch between how we ultimately deploy these systems and how we train them, leaving much of search and synthesis unoptimized.

We introduce SPIRAL, an RL framework for making all inference-compute primitives end-to-end learnable: models learn to coordinate sequential, parallel, and aggregative reasoning using only the reward of the final output.  Work with @ifdita_hasan (co-lead), @michaelyli_ , @oshaikh13 , @yoonholeee , @DorsaSadigh , @chelseabfinn , @noahdgoodman 🧵

🔗 View Quoted Tweet

💬5🔄10❤️77👀10881📊17

⚡ Powered by xgo.ing

@peterbhase @ChrisGPotts The post: https://t.co/Bn7ASHZ9Ap

Stanford AI Lab (@StanfordAILab) — Fri, 19 Jun 2026 18:44:40 GMT

@peterbhase@ChrisGPotts The post: ai.stanford.edu/blog/cot-monit…

💬1🔄3❤️15👀6620📊5

⚡ Powered by xgo.ing

New SAIL Blog post: CoT Monitoring: Where Does a Hot Safety Problem Come From? @peterbhase and @C...

Stanford AI Lab (@StanfordAILab) — Fri, 19 Jun 2026 17:11:25 GMT

New SAIL Blog post: CoT Monitoring: Where Does a Hot Safety Problem Come From?  

@peterbhase and @ChrisGPotts trace the history of a big idea in AI Safety

💬5🔄1❤️49👀19313📊16

⚡ Powered by xgo.ing

Modern multimodal models aren't a single decode loop anymore; they're composite. M* is one runtime t...

Stanford AI Lab (@StanfordAILab) — Thu, 18 Jun 2026 21:52:34 GMT

Modern multimodal models aren't a single decode loop anymore; they're composite. M* is one runtime that serves them all, and it matches or beats every specialized system: up to 2.7× on omni TTS, 12.5× on world-model rollouts. Learn more here: ai.stanford.edu/blog/mstar/

💬4🔄13❤️62👀11099📊17

⚡ Powered by xgo.ing

https://t.co/7hsAQFD68G

Stanford AI Lab (@StanfordAILab) — Wed, 17 Jun 2026 18:31:17 GMT

venturebeat.com/orchestration/…

💬0🔄0❤️3👀2959📊1

⚡ Powered by xgo.ing

Learn about how to orchestrate agents without a central orchestrator… in @VentureBeat’s recent artic...

Stanford AI Lab (@StanfordAILab) — Wed, 17 Jun 2026 18:30:53 GMT

Learn about how to orchestrate agents without a central orchestrator… in @VentureBeat’s recent article on DeLM! By @Mao_Yuzhen and @Azaliamirh

Azalia Mirhoseini@Azaliamirh

Thank you @VentureBeat for covering our work on Decentralized Language Models (DeLM)!

DeLM makes agentic tasks like coding and multi-doc Q&A more accurate and significantly cheaper, e.g. a ~10% jump on SWE-bench Verified with Gemini-3 Flash at less than half the cost.

yuzhenmao.github.io/DeLM/@Mao_Yuzhen

🔗 View Quoted Tweet

💬4🔄8❤️30👀13780📊10

⚡ Powered by xgo.ing

@jyangballin @vincentsunnchen https://t.co/P8vGVpGyc5

Stanford AI Lab (@StanfordAILab) — Wed, 03 Jun 2026 22:18:39 GMT

@jyangballin@vincentsunnchenyoutube.com/watch?v=2Sxaeu…

💬1🔄3❤️7👀7173📊4

⚡ Powered by xgo.ing

Hear from @jyangballin on ProgramBench and the lineage of AI coding benchmarks! In conversation with...

Stanford AI Lab (@StanfordAILab) — Wed, 03 Jun 2026 22:18:29 GMT

Hear from @jyangballin on ProgramBench and the lineage of AI coding benchmarks! In conversation with @vincentsunnchen

John Yang@jyangballin

Had the best time chatting w/ @vincentsunnchen! Shared some thoughts around how far we've come and how far we might go in AI coding land. Hope you enjoy!

🔗 View Quoted Tweet

💬3🔄5❤️24👀10611📊5

⚡ Powered by xgo.ing

Check out our latest blog post highlighting SAIL papers appearing at #CVPR2026. Congrats to everyone...

Stanford AI Lab (@StanfordAILab) — Wed, 03 Jun 2026 17:36:19 GMT

Check out our latest blog post highlighting SAIL papers appearing at #CVPR2026. Congrats to everyone! ai.stanford.edu/blog/cvpr-2026/

💬3🔄1❤️11👀3285📊4

⚡ Powered by xgo.ing

Read the full piece at https://t.co/uloN888wwZ

Stanford AI Lab (@StanfordAILab) — Fri, 29 May 2026 18:42:02 GMT

Read the full piece at foreignaffairs.com/china/chinas-a…

Avanika Narayan@Avanika15

Honored to be published in @ForeignAffairs.

For 2 years we've made the case for local AI. Here's the geopolitical angle: the U.S.–China race isn't about who trains the best model. It's about whose models, chips, and frameworks run by default on billions of devices.

Joint work w/@jdunnmon & @JonSaadFalcon 

(1/N)

🔗 View Quoted Tweet

💬1🔄6❤️10👀7096📊3

⚡ Powered by xgo.ing

Are you at ICLR 2026 in Rio? Check out the full list of papers from Stanford AI Lab - covering LLM r...

Stanford AI Lab (@StanfordAILab) — Thu, 23 Apr 2026 07:26:58 GMT

Are you at ICLR 2026 in Rio? Check out the full list of papers from Stanford AI Lab - covering LLM reasoning, agentic systems, AI safety, robotics, spatial intelligence, video generation, and more. See you there! 🇧�ai.stanford.edu/blog/iclr-2026NW

💬2🔄12❤️80👀15212📊15

⚡ Powered by xgo.ing

Check out our latest SAIL blog post on VAGEN, a reinforcement learning framework that trains VLM age...

Stanford AI Lab (@StanfordAILab) — Mon, 09 Mar 2026 22:10:59 GMT

Check out our latest SAIL blog post on VAGEN, a reinforcement learning framework that trains VLM agents to build internal world models through explicit visual state reasoning! ai.stanford.edu/blog/vagen/

💬4🔄15❤️91👀15117📊26

⚡ Powered by xgo.ing

Can AI actively explore and build mental maps of space, or just answer when handed observations? C...

Stanford AI Lab (@StanfordAILab) — Thu, 26 Feb 2026 00:28:14 GMT

Can AI actively explore and build mental maps of space, or just answer when handed observations?

 Check out our latest SAIL blog post on Theory of Space, a new benchmark probing whether foundation models can construct, revise, and exploit spatial beliefs through active exploration! 

ai.stanford.edu/blog/tos_stanf…

💬6🔄15❤️107👀12252📊29

⚡ Powered by xgo.ing

Collaboration between Stanford SAIL and ETH shows RL with rich feedback significantly outperforms sc...

Stanford AI Lab (@StanfordAILab) — Thu, 29 Jan 2026 20:47:48 GMT

Collaboration between Stanford SAIL and ETH shows RL with rich feedback significantly outperforms scalar rewards on very hard tasks!

Carlos Guestrin@guestrin

With SDPO, you can now do RL with natural language feedback, like error messages from coding environments or LLMs as judges. You can achieve huge gains over GRPO with scalar rewards!

🔗 View Quoted Tweet

💬4🔄10❤️88👀18447📊19

⚡ Powered by xgo.ing

Everyone can now discover new SOTA in science with a few hundred bucks! Test-Time Training + open m...

Stanford AI Lab (@StanfordAILab) — Thu, 22 Jan 2026 21:46:06 GMT

Everyone can now discover new SOTA in science with a few hundred bucks!

Test-Time Training + open model > prompt engineering + closed frontier model (Gemini, GPT-5), for discovery problems in Mathematics, Kernel Engineering, Algorithms and Biology.

https://t.co/IIN67Abc9H

Mert Yuksekgonul@mertyuksekgonul

How to get AI to make discoveries on open scientific problems?

Most methods just improve the prompt with more attempts. But the AI itself doesn't improve.

With test-time training, AI can continue to learn on the problem it’s trying to solve: test-time-training.github.io/discover.pdf

🔗 View Quoted Tweet

💬13🔄34❤️222👀37429📊56

⚡ Powered by xgo.ing

Our latest research, TTT-E2E, marks a new era for LLM memory. Now, models can continue training du...

Stanford AI Lab (@StanfordAILab) — Mon, 12 Jan 2026 19:07:35 GMT

Our latest research, TTT-E2E, marks a new era for LLM memory.
 
Now, models can continue training during deployment, using context as training data to update their weights and learn from massive amounts of experience. 

With @NVIDIAAI and @AsteraInstitute

Karan Dalal@karansdalal

LLM memory is considered one of the hardest problems in AI.

All we have today are endless hacks and workarounds. But the root solution has always been right in front of us.

Next-token prediction is already an effective compressor. We don’t need a radical new architecture. The missing piece is to continue training the model at test-time, using context as training data.

Our full release of End-to-End Test-Time Training (TTT-E2E) with @NVIDIAAI, @AsteraInstitute, and @StanfordAILab is now available.

Blog: nvda.ws/4syfyMN
Arxiv: arxiv.org/abs/2512.23675

This has been over a year in the making with @arnuvtandon and an incredible team.

🔗 View Quoted Tweet

💬8🔄30❤️164👀23576📊35

⚡ Powered by xgo.ing

SAIL is still accepting applications for the SAIL Postdoctoral Fellowships! This is an opportunity t...

Stanford AI Lab (@StanfordAILab) — Thu, 11 Dec 2025 20:30:05 GMT

SAIL is still accepting applications for the SAIL Postdoctoral Fellowships! This is an opportunity to work with our wonderful professors and community. Applications received by the end of December 15 will receive full consideration: ai.stanford.edu/postdoctoralfe…

💬5🔄25❤️180👀58280📊25

⚡ Powered by xgo.ing

Millions of children face speech disorders—but few get timely care. Our new benchmark, SLP-Helm, tes...

Stanford AI Lab (@StanfordAILab) — Tue, 28 Oct 2025 23:48:04 GMT

Millions of children face speech disorders—but few get timely care. Our new benchmark, SLP-Helm, tests how AI models diagnose pediatric speech—revealing promises, pitfalls, and bias. 

Read more on our blog w/ @sangttruong, @nickhaber, @sanmikoyejo, @stai_researchai.stanford.edu/blog/slp-helm/

💬1🔄10❤️19👀8167📊6

⚡ Powered by xgo.ing

We do a lot of cutting edge research at the Stanford AI Lab, but really our main job is educating st...

Stanford AI Lab (@StanfordAILab) — Tue, 25 Mar 2025 01:38:03 GMT

We do a lot of cutting edge research at the Stanford AI Lab, but really our main job is educating students.  Here is a list of great SAIL Graduates of 2025, who are variously looking for academic and industry jobs! 💪 
ai.stanford.edu/blog/sail-grad…i

Compiled by @NikilSelvam Alex Nam @judyhshen

💬10🔄50❤️90👀12395📊32

⚡ Powered by xgo.ing

Check out our latest blog post about MiniVLA, a smaller open-source vision-language-action model! ht...

Stanford AI Lab (@StanfordAILab) — Sun, 15 Dec 2024 18:19:05 GMT

Check out our latest blog post about MiniVLA, a smaller open-source vision-language-action model!
ai.stanford.edu/blog/minivla/

💬5🔄11❤️83👀12878📊22

⚡ Powered by xgo.ing

Congratulations to @GuibasLeonidas and colleagues for winning the SIGGRAPH 2023 Test of Time award f...

Stanford AI Lab (@StanfordAILab) — Tue, 08 Aug 2023 20:19:06 GMT

Congratulations to @GuibasLeonidas and colleagues for winning the SIGGRAPH 2023 Test of Time award for their paper “Functional Maps: A Flexible Representation of Maps Between Shapes”!

dl.acm.org/doi/10.1145/21…people.csail.mit.edu/jsolomon/asset…

💬0🔄2❤️16👀5174📊2