Thomas

[Thomas @tjomiii]

@ChatGPT @googledrive @grok how does this work. Is there an api to put Google Docs inside ur app?

[Jitendra MALIK @JitendraMalikCV]

I see every week on X an announcement or demo which implies that robotic manipulation has been solved. The only reason I don't believe it is because manipulation had already been solved last week by somebody else! So may I propose the "5 year old paired comparison test" ? At the next conference let's set up a number of tables to which you can bring your robot hardware. Next to it we will have another table where there will be a 5 year old child. In parallel we will try 100 different manipulation tasks that a neutral person has chosen- we could start with "pick up anything" - only household objects (e.g. as might be found in a typical American home) will be used, and we compare the performance of your robot with that of the 5 year old. Can you pick up a coin? Or a book? Or untwist a bottle top? Or insert any plug into a matching socket? Rotate one face of a Rubik's cube? Until your robot can do all the "open world manipulation" that a 5 year old kid can, some humility is in order.

[Jonathan (Buzz) Embley-Riches @JonEmbleyRiches]

Releasing the alpha of Unreal Robotics Lab — an open-source Unreal Engine plugin with full MuJoCo physics.

Photorealistic rendering and accurate contact physics. No compromises on either side.

GitHub: https://t.co/GAvwsncqtG
Paper: https://t.co/w8sbILR7dW

[Chetan Puttagunta @chetanp]

A truly sensational product and company. Legora has gone $1M to $100M ARR in less than 18 months. Setting a new speed record for the fastest ever to do it with a direct sales motion. Thrilled to have backed this remarkable group of people at seed, 7 months before they launched.

[Thomas @tjomiii]

@HaoruXue @DrJimFan @yukez @TairanHe99 @zhengyiluo @eric_srchen @ziang_cao @80gg_overmind @TingwuWang @ruijie_zheng12 @fengyuan_hu Lads

[Robert Youssef @rryssf]

RIP fine-tuning ☠️

This new Stanford paper just killed it.

It’s called 'Agentic Context Engineering (ACE)' and it proves you can make models smarter without touching a single weight.

Instead of retraining, ACE evolves the context itself.

The model writes, reflects, and edits its own prompt over and over until it becomes a self-improving system.

Think of it like the model keeping a growing notebook of what works.
Each failure becomes a strategy. Each success becomes a rule.

The results are absurd:

+10.6% better than GPT-4–powered agents on AppWorld.
+8.6% on finance reasoning.
86.9% lower cost and latency.
No labels. Just feedback.

Everyone’s been obsessed with “short, clean” prompts.

ACE flips that. It builds long, detailed evolving playbooks that never forget. And it works because LLMs don’t want simplicity, they want *context density.

If this scales, the next generation of AI won’t be “fine-tuned.”
It’ll be self-tuned.

We’re entering the era of living prompts.

[Kyle🤖🚀🦭 @KyleMorgenstein]

sim + RL is going to eat all the teleop + IL work alive once we can figure out how to generate diverse enough data. you just can’t beat the scale, and we know how to do sim2real now.

[Haoru Xue @HaoruXue]

🚀 Introducing LeVERB, the first 𝗹𝗮𝘁𝗲𝗻𝘁 𝘄𝗵𝗼𝗹𝗲-𝗯𝗼𝗱𝘆 𝗵𝘂𝗺𝗮𝗻𝗼𝗶𝗱 𝗩𝗟𝗔 (upper- & lower-body), trained on sim data and zero-shot deployed.

Addressing interactive tasks: navigation, sitting, locomotion with verbal instruction. 🧵

https://t.co/LagyYCobiD

[Dvij Kalaria @DvijKalaria]

1/6❓Can robots learn to play competitive sports by observing expert human players?
🚨Introducing LATTE-MV, a scalable 3D reconstruction system that⚙️800+ hrs of Youtube videos to create the largest 3D 🏓TT dataset!
✅27 hrs of gameplay | 73k exchanges

🌐https://t.co/55uNyHu6yE

[Thomas @tjomiii]

@mikkelsvartveit 🦙🦙🦙🦙🦙

[Thomas @tjomiii]

@alex_prompter Nothing about these prompts are critical, can we stop sounding like ChatGPT!!!

[Thomas @tjomiii]

@NephTalks @099h057 @Mmaaarijuana @elonmusk Epic rap battle guys! But honestly, why so mad? Think this mess of a political situation would be a lot better if both sides were able to discuss and disagree in a more composed and constructive manner 👯

[Mario Laul @mlphresearch]

"Dee maintained that if you give computer people more time, they will just consume it." So he always insisted... on shorter projects with uncompromising deadlines.

https://t.co/L0amYh0LFC

[Yann LeCun @ylecun]

Nice.

[Yann LeCun @ylecun]

Claims that the US might become Nazi Germany if Trump is elected may seem a little far-fetched (though the analogy is striking), but not nearly as far-fetched as @elonmusk's claims that the US nay turn into Venezuela if @KamalaHarris is elected.

I'd say it's more likely that the US will become just a little bit more like Norway: very high income, universal healthcare, free public higher education, lots of oil *and* lots of Teslas.

[Sanchit Gandhi @sanchitgandhi99]

Introducing the smallest Distil-Whisper model yet!

distil-small.en is over 10x smaller, 5x faster and within 3% WER of large-v2 🎯

At just 166M parameters, it's is perfect for low-memory environments, such as on-device or mobile 📞

Get started here: https://t.co/G25NIBDJ4i

[Thomas @tjomiii]

@mikkelsvartveit Thx mom!

[Sam Altman @sama]

@elonmusk anytime 🫡

[Danby Choi @danbychoi]

Hvordan kan Beyoncés karriere blomstre gjennom de siste tiårene hvis kvinner og svarte er så utfordret som et utall kronikker ledet oss til å tro?

Hvorfor lever jeg hvis selvmord liksom er et problem?

Og vennligst forklar hvordan det kan være global oppvarming all den tid jeg sitter her og fryser?

Les Natt og dag for å bli dummere:

[Chris Bakke @ChrisJBakke]

Update:

Microsoft and OpenAI execs sent Sam Altman a Microsoft Teams link, which he refuses to use.

Sam Altman reciprocated by sending a Google Meet link for the call, which Satya refuses to use.

Currently at a stalemate.