J'ai parlé de lancer des éclairs à un de mes enfants. Il m'a dit que c'était impossible en vrai.
Je lui ai montré une vidéo du Palais de la Découverte.
Sauvons le Palais !
www.change.org/p/sauvons-le...
@srequena.bsky.social
CTO at Genci the 🇫🇷 #HPC agency deploying supercomputers for numerical #simulation, #AI and #quantum computing
J'ai parlé de lancer des éclairs à un de mes enfants. Il m'a dit que c'était impossible en vrai.
Je lui ai montré une vidéo du Palais de la Découverte.
Sauvons le Palais !
www.change.org/p/sauvons-le...
#Supercalculateur #JeanZay : la France quadruple ses ressources en calcul scientifique, et fait fois 13 pour l’ #IA
👉 Retour sur l'inauguration de ce nouvel outil de pointe au service de la #science et de l' #innovation
@srequena.bsky.social @cnrs.fr @agencerecherche.bsky.social
👉 urls.fr/6_5b2t
#AIFactory par @srequena.bsky.social de GENCI dans l'Usine Nouvelle
💡 🇪🇺 "Face aux besoins croissants en capacités de calcul, portés notamment par l’essor de l' #IA générative, l’Europe a annoncé la création de 13 AI Factories... 🇫🇷 La France aura la sienne"
Pour en savoir plus 👇
lnkd.in/ebYsZz83
Congrats Nice event 👏🏻👏🏻👏🏻
04.04.2025 17:18 — 👍 1 🔁 0 💬 1 📌 0#BuyEuropean when possible 🇪🇺
03.04.2025 18:44 — 👍 49 🔁 8 💬 0 📌 1Se tient aujourd’hui la Journée mondiale de sensibilisation à l' #autisme #SantéMentale. À cette occasion, retour sur un projet de chercheurs au sein du centre #NeuroSpin, ayant recours au #supercalculateur Jean Zay @srequena.bsky.social @cea.fr www.linkedin.com/posts/genci_...
02.04.2025 12:10 — 👍 2 🔁 1 💬 0 📌 0SemiAnalysis sort le 1er comparatif indenpendant nommé ClusterMAX des NeoClouders pour #IA et bonne nouvelle @scaleway.com est très bien classé 👏🏻👏🏻👏🏻 #AIFactoryFrance
semianalysis.com/2025/03/26/t...
News Flash! ⚡
The @eurohpc-ju.bsky.social has selected #six new #AIFactories to strengthen Europe's #AI leadership! 🌍
Austria, Bulgaria, France, Germany, Poland, and Slovenia will host cutting-edge AI hubs, to boost #European AI innovation.
Read more: europa.eu/!rP6D6M
Fermeture du bureau de la scientifique en chef à la Nasa et autres nouvelles réductions pour l'agence spatiale américaine : www.cieletespace.fr/actualites/l...
11.03.2025 17:20 — 👍 4 🔁 5 💬 0 📌 0C est fait 👍 bravo @firefly-aerospace.bsky.social 👏🏻👏🏻👏🏻
02.03.2025 08:38 — 👍 2 🔁 0 💬 0 📌 0US climate researchers come in Europe to make your research great again 🚀
22.02.2025 11:08 — 👍 3 🔁 0 💬 0 📌 0The best plane ever made 🤩
19.02.2025 06:59 — 👍 0 🔁 0 💬 1 📌 0Congrats 👏🏻👏🏻👏🏻 fully deserved
13.02.2025 06:19 — 👍 1 🔁 0 💬 0 📌 0Nominations open for the Jack Dongarra Early Career Award
Researchers with a career 4-10 years after their Ph.D. are eligible
The award includes a certificate of recognition and a cash prize of 5,000 Euros
Nomination Deadline: Feb 26, 2025
isc-hpc.com/jack-dongarr...
#HPC #ISC25
👏🏻👏🏻👏🏻
21.01.2025 21:03 — 👍 1 🔁 0 💬 0 📌 0Welcome on @bsky.app colleagues 👍
21.01.2025 20:57 — 👍 0 🔁 0 💬 0 📌 0👏🏻 @deepseek.bsky.social DeepseekV3 is out with SOTA performance beating Claude, Llama 405b etc on 10x less compute, $6m worth (2.8m hours) on 2048 H800s (cut down H100s) and trained using fp8 (quite rare even now on big LLM training)
github.com/deepseek-ai/...
Was always my surprise when coming in the US, the so big heterogeneity of quality of states power grids
Also what are the plans of Texas for fatal heat reuse, the photo shows a data center lost in the middle of nowhere !
A gigantic step forward
First demonstration of quantum teleportation over busy Internet cables
Advance opens door for secure quantum applications without specialized infrastructure
news.northwestern.edu/stories/2024...
2025 will be the year of AI for science
Leveraging all the things we've recently learned training AI models for 1000x impact in science
and this will need data!
More details: huggingface.co/blog/lemater...
Thread:
...
Microsoft has just announced Phi-4.
Learn about Phi-4, the latest small language model in Phi family, that offers high quality results at a small size (14B parameters).
Phi-4 outperforms larger models on math related reasoning due to advancements.
techcommunity.microsoft.com/blog/aiplatf...
University of Salamanca in Spain in 1218
University of Toulouse in france in 1229…
The FineWeb team is happy to finally release "FineWeb2" 🥂🥳
FineWeb 2 extends the data driven approach to pre-training dataset design that was introduced in FineWeb 1 to now covers 1893 languages/scripts
Details: huggingface.co/datasets/Hug...
A detailed open-science tech report is coming soon
Became my wallpaper ! 🙏
28.11.2024 21:07 — 👍 0 🔁 0 💬 0 📌 0many strange sentences but yes HPL is good for stressing components at full energy levels but better play with real apps (unitary & workload) and also synthetic good ones for some given features
Leading benchmarks are so tuned by vendors that they are irrelevant on real life, also no word on HPCG 🥲
@hephtaicie.bsky.social you make my day 😂
26.11.2024 18:10 — 👍 1 🔁 0 💬 0 📌 0Amazon, Apple and NVIDIA compete for #GPU/accelerator talent with salaries (in some cases) of $300k or more
www.rdworldonline.com/amazon-apple...
#AI #HPC
#SC24 Panel: #HPC Meets #QuantumComputing
When and How Will Applications Benefit?
And Which Ones?
o PNNL
o LRZ
o GENCI
o TUM
Applications that specifically leverage both HPC & QC resources together are still a rarity
Ongoing activity and progress is remarkable
www.hpcwire.com/2024/11/21/w...