@ph_singer Congratz guys and proves reality in Germany is much more positive than the negative spin the news often pushes.
English
Markus.JM
260 posts
Markus.JM
@MarkusJm
New acc, former acc: Maiernator https://t.co/LFrmEU9uW4 https://t.co/QozaoTEKGU
Germany Katılım Ekim 2025
103 Takip Edilen6 Takipçiler
So excited about this! We are also continuously looking for excellent talent joining our independent frontier AI lab, please get in touch!
Prior Labs@prior_labs
It's official: Prior Labs has closed its acquisition by @SAP. We continue as an independent frontier AI lab - team and mission intact, open by default - backed by €1B+ to build the future of AI for enterprise data. We're only getting started: priorlabs.ai/blog-posts/pri…
English
End of my world cup period.
Spain was stronger than France. The referee helped Spain, but still, France never was dangerous.
I'll move back to AI/ML focus end of week. Now I'm going to enjoy the rest of my week off.
English
More info: Prompt your ChatGPT the following way:
“Compare the performance of the US, Canadian, UK and the German health care systems along the following metric: survival time conditional on getting certain diseases.”
You might be surprised.
Rudi Bachmann@BachmannRudi
He is not answering the original question by Levin. I hope everyone understands that.
English
@florianederer You could write him you have money to offer as you are a nigerian prince.
English
Exciting business proposition. I think I'm going to leave academia.
English
I haven't had Claude Fable successfully complete a single prompt yet.
English
All these players are real French men.
Not sure why this racist guy appears on my TL, but I'd love to not see morons like that.
𝐀𝐍𝐓𝐔𝐍𝐄𝐒@Antunes1
How can any real Frenchman feel any emotional connection to this team?
English
That's precisely why I shifted from math to AI for my PhD. I was working on algebraic topology. The cartoon is spot on.
I didn't want to work on something I could only discuss with a dozen people worldwide.
I did not expect Ai to become something discussed by millions either.
Mathieu@miniapeur
English
Starting on ARC AGI3 comp on @kaggle. This looks extremely challenging.
I am not discussing ARC AGI3 in general where one can use frontier LLMs and unlimited time. I am discussing Kaggle specific with one GPU, not internet access, and limited time.
Current top score is about 1.20%. And I'm sure it is not easy to get to that.
English
@JFPuget @C_A_G0101 Science is difficult for many and to cite my favorite comics.
Nous n'avons peur que d'une chose : c'est que le ciel nous tombe sur la tête.
Français
@C_A_G0101 Merci, je me souvenais de cette sequence et je voulais la chercher.
Français
People asked me for an example of the limit I see in current coding agents.
I am implementing a distributed RL environment. One part of rollout is very irregular and does not run fast on GPU. Docex decided to run it on cpu. So far so good.
Given the large number of environments, it decided to run the code for each env in parallel using a python multi processing pool. Still good so far.
Yet the code crashed the cloud instance I was using.
I found that it sized the pool by looking at the number NC of available cpu. The issue is that in a distributed setting with N workers, each worker would spawn NC processes, creating a total of NxNC processes. Each cpu is oversubscribed N times. If the processes are heavy enough this crashes the machine.
This is what I mean by good local view but lack of system wide view. There is no model of how the code execution behaves.
Once we understand that limit then we can use coding agents more effectively.
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱@JFPuget
Total BS if you ask me. Maybe it is because I don't use chatbots, but I don't get how one could think for a second what Hinton says. I use coding agents (codex currently), and the amount of times I have to correct it keep me immune from the delusion Hinton seems to have. I use them because they accelerate me. But I have to steer them closely. It is like drawing a car until recently (before cars can self drive).
English
Wenn die AfD wirklich Wahlen gewinnen wollte, böte sie Führungen durchs Frankfurter Bahnhofsviertel an. Ein sichtbareres Politikversagen gibt es in Deutschland derzeit nicht.
Deutsch
SPACEX valuation shows that markets value narratives over reality.
Elmo Has been smart capitalizing on cult like behavior of investors.
I do think that reality wins in the end. This will be interesting to watch.
English
Oh look, AI folks have discovered stacking!
OpenRouter@OpenRouter
Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇
English
Before we did it, no other country dared have two people pummel each other for entertainment.
Acyn@Acyn
Rubio: President Kennedy announced that we were going to put a man on the moon. We did it. We are a nation founded on doing what no one else dared to do. And at some level, that's what this whole company, what UFC has been
English
@saltyAom Glad some are opening eyes in the US. That was clear form outside for quite a while already.
English
If a government can dictate your company what or what not to do then how is it any difference from CCP at all
Anthropic@AnthropicAI
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
English
@ben_golub @MeganTStevenson @RefineDotInk For none graphic output that should be quite trivial, comparing the output of the code with those in the tables (as long as the code is in R or Python and not SPSS,SAS). For graphs there might be huge gaps in interpretation of an agent vs whats in the paper.
English
@MeganTStevenson @RefineDotInk Yeah that is on our roadmap and seems quite important to me
English
Has anyone built an AI tool that compares what a paper says it does to what is actually done in the code? Would love to see that out there, both as a way to double-check my own projects and a way to review published papers with code repositories.
English
name a single country where Muslims coexist peacefully with other faiths?
English
@darrell1114 @MoMohler It does not matter, If someone is not talented playing Football, he is not talented, simple as that. Jokic played since childhood, he was not even close of becoming pro.
Many athletic kids try it in their youth and suck at it...
English
