![Daniel J. Beutel Profile](https://pbs.twimg.com/profile_images/1311639774445416453/pixnHAiN_x96.jpg)
Daniel J. Beutel
@daniel_janes
Followers
368
Following
1K
Statuses
632
Train different @flwrlabs
Hamburg, Germany
Joined December 2013
RT @MattHartman: Submit a talk to the @flwrlabs 2025 AI Summit in March to showcase what’s possible with decentralized AI. Share your proj…
0
3
0
RT @gingsmith: Are you a junior researcher interested in federated and collaborative learning? Consider applying to the Simons Institute Re…
0
12
0
RT @ivelini: Flower v1.15 is out! Includes @LeRobotHF example via @flwrlabs If you were waiting for an official release to start building…
0
4
0
Boom XB-1 would be a great name for a new foundation model Kidding aside, amazing work by a small team. Congrats @boomaero !
0
0
0
Hear the latest from our friends at Intel!
🚀 Get ready for Patrick Foley @intel at FS25: He has a big suprise ready for the Flower community! 🌍 We’re excited to announce that Patrick Foley, OpenFL Lead Architect and Engineering Manager at @IntelSoftware, will be speaking at the Flower AI Summit 2025 -- and showing a demo of interest to everyone in the world of federated learning (FL)! 🎙️ Patrick has been at the forefront of advancing FL for many years now. Through his leadership in the OpenFL project, he has in particular championed the creation of federated standards: an initiative designed to foster interoperability and collaboration across FL frameworks and more! 🎫 Join us at King’s House, London, or online, March 26–27, 2025. Find the registration link in the comments below
0
0
1
AI desktop apps will converge to the same tech stack most other desktop apps use these days: a web app wrapped using Electron. Models will run on WebGPU once the software stack matures.
.@Jordan agree but curious where web GPU could fit in, if at all - too clunky?
0
0
1
Incredible to see the DeepSeek story unfold in the media. Little inaccuracies here and there have now morphed into just plain wrong statements. "DeepSeek is more efficient to run" --wrong, it's more efficient to train "Nvidia GPUs are no longer required" --wrong, DeepSeek still trains on Nvidia GPUs
2
0
8
A big factor leading to the market share difference between AMD and Nvidia is software, not hardware The entire ML ecosystem is Nvidia-first, a hard MOAT to overcome
Shouldn’t DeepSeek paper be boon to AMD? H800 are nerfed with half the interconnect speed compared to H100 but the DeepSeek team proved you don’t need all that bandwidth anyway because you can almost perfectly overlap comms with computation with few creative techniques!
0
0
1
RT @garrytan: This seems like an overreaction Wall Street needs to read the Wikipedia page on Jevon’s Paradox
0
351
0
Even football (or soccer as some people call it) teams are recruiting AI researchers now Which domain is not going to be affected by AI? Hard to come up with examples
We’re hiring a Research Engineer at @Arsenal ⚽️ to help accelerate our applied research efforts! We’re building state-of-the-art models to understand football at scale, and a suite of ML-powered applications on top of them to drive tactical and strategic decisions.
0
0
1
RT @flwrlabs: Flower AI Summit 2025 is coming in hot 🚀 Join us at the world's premier event for decentralized AI innovation; whether you’r…
0
6
0
@flwrlabs We'll have a few surprises in store for in-person attendees. Better register soon, there aren't that many on-site spots left.
0
0
4
RT @RemiCadene: We are not so far from a future where robots will be constantly learning by interacting with humans and their environments.…
0
25
0
@flwrlabs What a nice contribution! @LeRobotHF + @flwrlabs work really well together. There are many projects begging to be built...
0
0
4