The Top AI Papers of the Week (June 7 – June 14) – Agentopia
– Self-Harness
– Agents' Last Exam
– MiniMax Sparse Attention
– Lookahead Sparse Attention
– How AI Agents Reshape Knowledge Work
– The Geometry of On-Policy Distillation Read on for more:
MACHINE LEARNING
-
Top AI Papers of the Week June 7-14
By
–
-
DataMaster: an autonomous agent that processes AI data
By
–
Your AI model is stuck because of bad data, and not bad algorithms? Researchers from Shanghai Jiao Tong University, Carnegie Mellon University and other leading institutions present DataMaster: an autonomous agent that processes
-
Siemens AI-driven factory project with Nvidia at Hannover Messe
By
–
At Hannover Messe, Ujjwal Kumar of Siemens Digital Industries told me about their fully AI driven factory project with Nvidia in Erlangen. Partner content with @Siemens. #sie_us pic.twitter.com/uJywl2UyyH
— Lucian Fogoros (@fogoros) 14 juin 2026At Hannover Messe, Ujjwal Kumar of Siemens Digital Industries told me about their fully AI driven factory project with Nvidia in Erlangen. Partner content with @Siemens
. #sie_us -
LeCun’s new paper proves AI world model accuracy with LeJEPA
By
–
Yann LeCun's new paper just proved when AI truly learns the world. Most AI systems learn an internal picture of the world. Nobody could prove that picture is correct. LeCun's new research finally provides that proof. It studies LeJEPA, a method that trains models to predict
-
A bracelet turns hand gestures into robot training data
By
–
A bracelet that transforms hand gestures into training data for robots https://
youtu.be/nUFPYMUBK6o?si
=JvuhEQwe-QYNeLGA
… via @YouTube #humanoidtech #humanoid #robot #Robotics #AI #TechRevolution #TechInnovation #ArtificialInteligence #PhysicalAI @lexfridman @KirkDBorne @Ronald_vanLoon -
Massive scale difference between 7B and 3T parameter models
By
–
There is a MASSIVE difference in scale and engineering between a 7B model (which pretty much anyone can train and serve) and a 3T parameter model trained on 50T high-quality tokens across 10K to 50K B200s. It’s like comparing fireworks with a rocket that goes to the moon. While
-
@datachaz — 2026-06-14
By
–
Repo link → https://
github.com/rohitg00/ai-en
gineering-from-scratch
… Shoutout to @ghumare64 for architecting this whole thing and giving it away to the developer community for free. It takes an insane amount of effort to build something this comprehensive. Drop a on the repo and start building! -
Doubt on masked image modeling applied to JEPA
By
–
I had the impression that it was just masked image modeling applied to JEPA. I might be missing something though – what impressed you?
-
Did they unlock Claude Fable 5? Test with system prompts
By
–
ACABAN DE DESBLOQUEAR CLAUDE FABLE 5?
— Nico (@nicos_ai) 14 juin 2026
Anthropic publicó sus system prompts.
Así que alguien cogió el de Fable 5 y lo metió en Opus 4.8.
Mismo modelo en las dos pantallas.
Solo cambia el prompt.
Le pidieron a los dos lo mismo:
“crea una landing page estilo Apple”
→ Opus… https://t.co/dCycF8Zaf6 pic.twitter.com/YE8qxKqhgWHAVE THEY UNLOCKED CLAUDE FABLE 5? Anthropic has published its system prompts. So someone took the one from Fable 5 and injected it into Opus 4.8. Same model on both screens. Only the prompt changes. They were asked the same thing to both:
"create a landing page style"
