Jan Hünermann

Most likely training neural nets right now. Ex @wayve_ai, @poolsideai, NUS, TUM, CDTM

[Jan Hünermann @janhuenermann]

Great to see progress on LingoQA! The community came a long way from the 60% in our paper. Qwen 3.8-Max achieves 84.8% and is - to the best of my knowledge - the current SoTA. https://t.co/v2w1WABGDd

[Jensen Huang @JensenHuang]

Today, we’re launching Alpamayo 2 Super, our frontier open reasoning model for autonomous vehicles.

Beyond seeing, Alpamayo understands and reasons through the complex world - thinks before it acts.

It’s a powerful backbone for robotaxis, trucks, shuttles, delivery vans, tractors and the long tail of mobile robots—billions of autonomous machines someday.

We’re releasing it for commercial use under OpenMDW-1.1 so teams can inspect it, fine-tune it and deploy it—open models advance safety and security.

The next wave of AI is robotics—and it starts with autonomous vehicles.

Great work, Alpamayo team!

https://t.co/2PYCCXWjZh

[Jan Hünermann @janhuenermann]

@qiyang_li Great work. Did you notice any benefits of doing more distillation than TD learning steps on each iteration (or vice versa)?

[Jan Hünermann @janhuenermann]

Excited to be at @NeurIPSConf in San Diego next week! If you want to chat large scale RL, agents, synthetic data, reward hacking, or want to know more about what we’re up to @poolsideai, I’d love to connect/grab a coffee.

Our booth is #913, come say hi! ⛱️

[Jan Hünermann @janhuenermann]

It blows my mind that a 8B model is able to outperform GPT4o/Sonnet here.

[wh @nrehiew_]

They also evaluate on the distilled models and distillation really just works. They even beat Qwen's very own QwQ.

At 8B parameters, it is matching Sonnet and has surpassed GPT-4o

[Jan Hünermann @janhuenermann]

@kellerjordan0 @Grad62304977 Oh yeah that makes sense! So most of the gains come from having a shortcut at all and the additional shortcut makes it a tiny bit easier for the model to use. Thanks :)

[Jan Hünermann @janhuenermann]

The ultimate application of AI will be sorting your shopping list by supermarket layout.

[Jan Hünermann @janhuenermann]

LINGO-2 is the first vision-language-action model that can drive a car on public roads. Excited for the future work that this opens up!

[Wayve @wayve_ai]

🚙💬Meet Lingo-2, a groundbreaking AI model that navigates roads and narrates its journey. Watch this video taken from a LINGO-2 drive through Central London 🇬🇧 The same deep learning model generates real-time driving commentary and drives the car.

[Jan Hünermann @janhuenermann]

End to end AI is the way towards generally capable robots. Great progress on robot dexterity for real world applications!

[Tony Zhao @tonyzzhao]

Introducing 𝐀𝐋𝐎𝐇𝐀 𝐔𝐧𝐥𝐞𝐚𝐬𝐡𝐞𝐝 🌋 - Pushing the boundaries of dexterity with low-cost robots and AI. @GoogleDeepMind

Finally got to share some videos after a few months. Robots are fully autonomous filmed in one continuous shot. Enjoy!

[Jan Hünermann @janhuenermann]

Backdoor in liblzma through a mix of social engineering and clever obfuscation. It modifies the RSA decrypt function on a linux system, potentially allowing unauthorised access over ssh! https://t.co/zus2HwsbC7

[Jan Hünermann @janhuenermann]

@itsnamgyu @ylecun @AravSrinivas These are not constraints of the *autoregressive* model itself but of the (currently popular) training paradigm. You can imagine using RL to train the model to perform CoT. Compute efficiency doesn’t matter if it gives us AGI (take GPT-4 for instance).

[Jan Hünermann @janhuenermann]

@ylecun @AravSrinivas What about chain of thought prompting where the chain of thought would be considered the reasoning step?

[Jan Hünermann @janhuenermann]

New paper demonstrating large language models with 1bit weights (+/-) scale just as well as FP16.

This potentially opens up 100B+ params open-source models running on consumer hardware

https://t.co/G4dTnD00eZ

[Jan Hünermann @janhuenermann]

I’m in Paris this week for ICCV. Drop me a message if you want to meetup!

[Jan Hünermann @janhuenermann]

With LINGO-1 we have a built an AI system that explains its own actions in natural language. One step closer towards embodied AI that reasons about the world like humans 💬

[Alex Kendall @alexgkendall]

Language is the future for how we interact with robots.

Today @wayve_ai is sharing a first look at LINGO-1, a new vision-language-action AI model. To give you a glimpse of its capabilities, here is a video of me playing with LINGO-1 yesterday morning.

[Jan Hünermann @janhuenermann]

Super excited about the progress we've made with world models!

[Wayve @wayve_ai]

🚨 Wayve Presents: Generative Artificial Intelligence for Autonomy (GAIA-1). #GAIA1 is a new generative AI research model that creates realistic driving videos and offers fine-grained control over ego-vehicle behaviour and scene features. https://t.co/A1me9oVuMq #GAIA

[Bill Gates @BillGates]

I’m a car guy who enjoys the meditative (and fun) aspects of driving–but I’m also eagerly anticipating the day when I can hand over the keys to a machine. https://t.co/aMQjTLZVT1

[anton @abacaj]

LLaMA has been fine-tuned by stanford,

"We performed a blind pairwise comparison between text-davinci-003 and Alpaca 7B, and we found that these two models have very similar performance: Alpaca wins 90 versus 89 comparisons against text-davinci-003."

[Jan Hünermann @janhuenermann]

Another great example showing that robotics is mostly a software problem! The "Egg Peeling Challenge" could be considered the Turing Test for embodied AI.

[Chelsea Finn @chelseabfinn]

Can you peel an egg without tactile feedback? 🥚 Turns out yes!

We had some fun pushing the limits of our robot hardware last night. 😄🤖

[Jan Hünermann @janhuenermann]

Are you an autonomous driving or AI researcher? Wayve is organising this year's "Scalable Autonomous Driving" workshop at ICRA. Paper submission is open now until 20th of March. https://t.co/cB7pg3jzVo

[Oleg Sinavski @SinavskiOleg]

Excited to announce the #ICRA2023 workshop on Scalable #Autonomous Driving!
CFP: https://t.co/tnCFQ3qmfT (by 20th March). Autonomous Driving papers are all welcome (extended abstracts or full papers)! #AI #Robots #DeepLearning #computervision #machinelearning