Tim Rädsch

Stealth | TUM, Helmholtz, KIT |
Visual Intelligence, Physics, World Models, Benchmarking, Evals, AI and startups.

[Iman @imodarressi]

Sam Altman created the most ruthless model in the world

we built the biggest mafia-like game ever
> 250.000 humans playing with/against AI

we found:
@OpenAI → maniac LIARS
@spacexai → by far the most LOYAL
@AnthropicAI → the biggest SNITCH

we also added a flybrain ;)

[Daniel Ek @eldsjal]

30 minutes to start a company. Why is this not the new default in Europe?

[Stefan Bauer @stefanAbauer]

The last of a sequence of works joint with the Google Paradigms of Intelligence work investigating growing and looping in LLMs. TLRD: These models are better sequence models! For more information see the thread from first author Ferdinand below.

[Ferdinand Kapl @KaplFer]

Looped LLMs are everywhere right now! But why are they interesting? Do they model language differently than regular LLMs?

Grown and looped transformers are better sequence models!

With @kaitlinmaile & @oswaldjoh (@Google, Paradigms of Intelligence Team) and @stefanAbauer. 🧵

[Varun Nair @_varunnair]

We will see researchers build great data companies.

Data is a research problem

[Tim Rädsch @TimRaedsch]

@raphaelmosaic @alexandriainc lfg

[Raphael Mosaic @raphaelmosaic]

Enterprises which win the AGI race will build their own internal Mercor.

Until now there was no way to build internal human datalabs without surveilling your workforce.

Introducing @alexandriainc

[Amir Karimi @AMK6610]

Excited to share “Growing to See”: growing SSL vision models provide better representations with up to 25% fewer FLOPs!

Co-led with @KaplFer.

With @Google’s Paradigms of Intelligence team (@kaitlinmaile & @oswaldjoh), @andrea_dittadi, Karl H. Johansson & @stefanAbauer.

1/13

[Tim Rädsch @TimRaedsch]

@karinanguyen Agree and it requires skill and taste.

There is nothing easier to create than to create a poorly designed benchmark.

[Tim Rädsch @TimRaedsch]

Could not agree more!

We have so much great talent and should work right now.

In parallel we need better basic conditions.
First: EU-INC as demanded and not watered down.

A decisive moment for Europe
@andreasklinger @_FriedrichMerz @Markus_Soeder @euinc_petition

[Rowland Manthorpe @rowlsmanthorpe]

Jensen Huang has a message for Germany

[Tim Rocktäschel @_rockt]

If in a few years we are wondering why Europe's startup ecosystem is still lagging behind, we will look back at this moment in time.

[Andreas Klinger 🦾 @andreasklinger]

At this point the whole EU Inc negotiations are so lost in the details that everyone forgot the main goal: Creating a standard.

Basically, allow founders to raise quickly globally and use stock options to hire competitively.

Let’s say it quite clearly:
1) If EU Inc isn’t providing a standard it’s useless. We already got 30+ legal entities in Europe. Experienced founders use Delaware Inc already because of this.
2) If EU Inc doesn’t fix stock options Delaware Inc is still the better option.

I understand the there are complex details and people are worried about loopholes. But you can’t lose the actual goal of it because of it.

An EU Inc that doesn't provide a standard is pointless. Nobody will care about intricacies of compromises reached in political committees if the market can't use the outcome.

If you are in contact with your political leaders (prime ministers, presidents, et al) please reach out to them. Or to founders/investors in your network who might have that access.

🔥🇪🇺✌️

https://t.co/HaV0NMKCe7

[Tim Rädsch @TimRaedsch]

Great opportunity. Check it out!

[Thaddäus Wiedemer @thwiedemer]

Robert Geirhos and I are #hiring student researchers at @GoogleDeepMind for 2027. Express you interest here: https://t.co/v4TV3x46BI

- full-time for 6 months
- on-site in Zurich
- co-advised by Robert Geirhos and me
- continuing our work on video models for visual intelligence

[Thaddäus Wiedemer @thwiedemer]

Robert Geirhos and I are #hiring student researchers at @GoogleDeepMind for 2027. Express you interest here: https://t.co/v4TV3x46BI

- full-time for 6 months
- on-site in Zurich
- co-advised by Robert Geirhos and me
- continuing our work on video models for visual intelligence

[Tim Rädsch @TimRaedsch]

@SinkingStarGame is an absolute blast! Awesome job @Jonathan_Blow and the whole team. Might be my favorite release of the year.

[Tim Rädsch @TimRaedsch]

@arankomatsuzaki @Nicolas_Keller delivers that with German precision from Munich and you can use it today! -> Germany-as-a-Service (GaaS)

[Anates Labs @Anates_Labs]

@odysseyml takes the first place on Physics-IQ Verified in the video to video (v2v) category with Odyssey 3 Pro.

Full Leaderboard: https://t.co/HvkfSlIfsW

In the image to image (i2v) category we still have a head to head race with @bfl_ai.

[Vincent Pauline @vincentpaulinef]

🆕 Very happy to share:

“One for All, All for One: Coordinated Multi-Agent Diffusion Steering via Stochastic Optimal Control”

Work co-led with top-notch collaborators @BarbanoRiccardo and Runchang Li 🤝

We show how to teach pretrained diffusion models to work together: coordinating robot arms, human motion, navigation and image components.

With frozen pretrained weights! ❄️

📄 Paper: https://t.co/oEnykDO0tx

Grateful to an incredible team: George Webber, Alexander Denker, Željko Kereta, @stefanAbauer, Francisco Vargas and Esmeralda S. Whitammer. 🙌

Idea, theory, and some examples below 🧵👇

[Anjney Midha @AnjneyMidha]

hundreds of MI355x and B300 nodes

now live on https://t.co/t0uE0bekz3

subsidized for all .edu and .govs

it’s time to scale the frontier ecosystem

for the public benefit

[NVIDIA AI @NVIDIAAI]

Physis-Lang with NVIDIA Cosmos 3 Super and Nano rank #1 and #2 on the Physics-IQ Verified image-to-video leaderboard.

The benchmark tests how well generated videos follow real-world physics.

Full rankings: https://t.co/dzOtUi0Aae