AlphaEvolve is deeply disturbing for RL diehards like yours truly
Maybe midtrain + good search is all you need for AI for scientific innovation
And what an alpha move to keep it secret for a year
Congrats big G
MACHINE LEARNING
-
AlphaEvolve reveals RL limitations, midtrain search suffices innovation
By
–
-
The Golden Age of AI Research: 2015 Optimism and Fundamentals
By
–
it must have been more fun to be an grad student in AI and ML circa 2015, when there was an immense amount of optimism that if you understood something True about the world (e.g. fundamentals of lingustics or the human visual cortex) then you could use that knowledge to build
-
Lumesys: AI-Powered Energy Tech for Industrial Consumers
By
–
Hi Peter, building Lumesys: Energy Tech leveraging AI and ML for emerging high enenergy consumers industries.
-
AI detects early cognitive decline signals through ambient intelligence
By
–
Through ambient intelligence, AI and computational neuroscientist @eadeli
’s research offers a solution for catching early signals of cognitive decline before it's too late. Learn more about this work supported by @stanfordhai -
ResNets: The Most Cited Paper of the 21st Century
By
–
The most cited paper of the 21st century (so far!) Deep Residual Learning for Image Recognition
-
DeepMind Announces New AI, Ushering in Era of Self-Improving Systems
By
–
Demis Hassabis annonce leur dernière IA chez deepmind. Nous entrons dans une ère où les algorithmes optimisent d'autres algorithmes — l’ère des systèmes auto-améliorants vient de commencer.
-
Common Machine Learning Misconceptions Explained
By
–
What’s a common misconception about machine learning that you wish more people understood?
-
Yann LeCun skeptical of LLMs as future of AI, cites limitations
By
–
Yann LeCun ne croit pas que les LLM soient l’avenir de l’IA.
— VISION IA (@vision_ia) 15 mai 2025
Pour lui, c’est une impasse :
→ Pas de vraie compréhension du monde
→ Pas de mémoire persistante
→ Raisonnement trop simpliste
Il mise sur autre chose. Et si c’était lui qui voyait juste… cinq ans trop tôt ?… pic.twitter.com/ikGkZRj8fhYann LeCun ne croit pas que les LLM soient l’avenir de l’IA. Pour lui, c’est une impasse :
→ Pas de vraie compréhension du monde
→ Pas de mémoire persistante
→ Raisonnement trop simpliste Il mise sur autre chose. Et si c’était lui qui voyait juste… cinq ans trop tôt ? -
Deploying DeepSeek-R1 Distill: GCP Infrastructure Trade-offs
By
–
Excited to share insights from our latest guest post on Decoding ML by @iusztinpaul
! We walk through deploying DeepSeek‑R1 Distill on three infrastructure models—each with its own trade‑offs: 1. Google Cloud Platform (GCP) • Self‑managed VM with an NVIDIA L4 GPU
• Full -
Amazon’s Flawed Recommendation Algorithm: Kayak Paddles and Pickleball
By
–
So if you are shopping for the *kayaking* paddles, Amazon will also recommend pickleball paddles.