
Cette nouvelle IA fait du raisonnement sans pré-entraînement, et elle tourne sur juste2 GPU). Si c'est vrai, c’est un truc vraiment énorme. 40,3 % sur ARC-AGI. Elle tourne sur un grille-pain. Comment c’est possible ? Je vais me renseigner pour vous.

By
–

Cette nouvelle IA fait du raisonnement sans pré-entraînement, et elle tourne sur juste2 GPU). Si c'est vrai, c’est un truc vraiment énorme. 40,3 % sur ARC-AGI. Elle tourne sur un grille-pain. Comment c’est possible ? Je vais me renseigner pour vous.

By
–

Meta vient (encore) de débaucher 3 chercheurs de google deepmind qui ont conçu le modèle médaillé d’or aux olympiades de maths (imo) d'il y a quelques jours. Le Zuckerberg est inarrêtable Il va finir par recruter toute la terre entière dans son labo

By
–
letting models think longer can decrease, not improve, accuracy "Inverse Scaling in Test-Time Compute" this paper shows that increasing inference steps produces an “inverse‑scaling” effect, longer CoT amplify distraction, framing over‑fit, deductive focus loss & more
By
–
Top 12 types of databases to keep on your radar in 2025!#database #technology @lexfridman @KirkDBorne @Ronald_vanLoon @erikbryn @antgrasso @sallyeaves @Nicochan33 @HaroldSinnott @mvollmer1 @marcusborba @CatherineAdenle @vinod1975 @YuHelenYu @ingliguori @fogoros @ylecun… pic.twitter.com/bLuWJyOCd8
— Amitav Bhattacharjee (@bamitav) 23 juillet 2025
Top 12 types of databases to keep on your radar in 2025! #database #technology @lexfridman @KirkDBorne @Ronald_vanLoon @erikbryn @antgrasso @sallyeaves @Nicochan33 @HaroldSinnott @mvollmer1 @marcusborba @CatherineAdenle @vinod1975 @YuHelenYu @ingliguori @fogoros @ylecun

By
–
Sapient Intelligence released Hierarchical Reasoning Model Trained on just 1K examples with no CoT or pre-training, the 27M parameter model handles complex reasoning tasks including extreme-soduko and maze puzzles

By
–
Google DeepMind won its first official gold-medal standard at International Math Olympiad The company's advanced version of Gemini with Deep Think solved 5 out of 6 problem statements, scoring 35/42 Last year, it won silver

By
–
Alibaba's Qwen team made headlines with three breakthrough releases this week: —Qwen3 non-thinking update with SOTA performance among open models
—Qwen3-Coder, an agentic coding model that tops benchmarks
—Qwen Code open-source command-line coding tool

By
–
incredible work on alignment steganography from anthropic fellows i've been looking for a straussian explanation of why china keeps publishing open models out of the goodness of their hearts if you do stuff like use open models to, idk, clean *ahem* synthetically paraphrase
By
–
GenAI Literacy: What Is It, and How Should We Teach It? — Frameworks, Reviews, and Approaches.

By
–
Artificial Intelligence for #Cybersecurity: http://
amzn.to/4gOK3rD v/ @PacktPublishing ——
#AI #MachineLearning #DataScience #Infosec —— 𝒦𝑒𝓎 𝐹𝑒𝒶𝓉𝓊𝓇𝑒𝓈: Learn AI methods and see how they fit into cybersecurity Learn how to design solutions in cybersecurity that