adrienbufort.bsky.social

Effectivement, si seulement on avait anticipé les chocs pétroliers en devenant des acteurs de pointe dans l’énergie la plus abondante, la plus propre et la moins risquée du monde ! Et si seulement les donneurs de leçon ne se battaient pas contre depuis 70 ans !

it feels that the real difficulty of doing weather forecasting using deep learning / diffusion is really the long term coherence.

Setting up my first android app with weather forecast using deep learning: you can download it here: play.google.com/store/apps/d...

Extensive lecture of Eric W. Tramel (Nvidia) on synthetic data: "Synthetic pretraining is the way frontier models are built" scalable-ai.eecs.berkeley.edu/assets/lectu...

Bah c'est précisément ce qu'ils cherchent à faire sauter. Il y a pas d'incompréhension, mais une volonté de maintenir la domination des plateformes US en Europe et faire de l'ingérence via les algos, IA, etc. Quand on voit ce qu'ils ont imposé à Tiktok c'est de la pure mauvaise foi

Trois ans après l’invasion de l’Ukraine, et une énorme vague de viols, enlèvements, tortures et meurtres de civils qui se poursuivent encore chaque jours, ce genre d’annonces de demi-mesures timorées ne fait que montrer l’incompétence et la lâcheté des dirigeants européens face à Poutine.

La France est la cible prioritaire des actions informationnelles russes en Europe. Et pour lutter contre ces tentatives de déstabilisation politique, il ne faut pas seulement couper les accès mais agir sur les causes de ce ciblage... Sinon, en 2027, on aura un(e) Trumpiste à l'Elysée...

Bluesky doesn't penalize you for using links, tags, using key words like Patreon or commissions, and doesn't have a central algorithm on your following feed, which means all the power rests on you to help your favourite creators be seen!

🌧️ 2024 figure comme l’une des années les plus pluvieuses depuis 1959 ! Grâce aux données de @meteofrance.com, cette carte permet de mieux comprendre la disparité régionale de ces précipitations, à l’heure d’une multiplication des événements climatiques extrêmes. 👉 www.ign.fr/mag/ou-t-il-...

I've been comparing cursor vs VSC continue plugin : my advise is that autocomplete cursor have more element compare to continue but for chat I prefer VSC control (you can choose any model you want gemini / groq etc)

Supposedly “isolationist” president-elect of the US is threatening to annex Canada, invade Panamá and declare war on Denmark, while the president-unelect (Musk) is actively working to topple European democracies (UK, Germany, Spain…). Good thing they're not interventionists! 😑

it's two years after the release of chatGPT in nov 2022 and google is a the top of lm ranking : lmarena.ai My guess is that, as they have the hardware (TPU), they are in a good spot to dominate the LLM area, cost wise and performance wise.

Happy New year bluesky !

I am looking at papers that are doing meteo forecast with DL / ML and I am surprised at how much easy is it to do compare to classic numerical method (but only if you have the data)

I was trying to find a good flow of research papers info : huggingface daily paper seems to be a nice way of doing (although a little be too focus on LLMs stuff) huggingface.co/papers

With all those discussions around test time inference, why we don't reuse some old results like : "End-to-end Algorithm Synthesis with Recurrent Networks: Logical Extrapolation Without Overthinking" arxiv.org/abs/2202.05826

It's incredible how ML / science bsky skyrocketed ! I see research paper everywhere now :)

Do people have reference on how to optimize deep learning inference pipeline regarding the framework used ? Is pytorch so good we don't need to switch to rust (like using huggingface candle lib) ?

anyone know of a good list of follows for generative modelling researchers? diffusion models, normalizing flows, SBI?

I got some much less political proganda on bluesky than on twitter / X it's super refreshing