What if a robot could learn from its own successes, not just its failures? This research introduces SRPO, a method that lets vision-language-action models use their own best attempts as a reference to assign smarter rewards. This self-referential approach boosted a robot's
MACHINE LEARNING
-
Achieving Peak Human Performance Across Multiple Domains
By
–
What does it take to achieve the highest level of human performance? Across athletics, science, chess, and music @ScienceMagazine https://
science.org/doi/10.1126/sc
ience.adt7790
… -
Finance Transformation: Automation and AI Drive Future Growth
By
–
The Transformation of Finance: Automation and AI Hold the Key
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @Scobleizer @AndrewYNg @drfeifei @KirkDBorne @fchollet @rowancheung @antgrasso -
Genspark CEO says ‘just prompt it’ – Microsoft Office in trouble
By
–
Genspark's CEO says "just prompt it."
— Robert Scoble (@Scobleizer) 18 décembre 2025
My reaction? Microsoft Office is in trouble as more people discover the new way to work. They won't be going back. https://t.co/cMltEAB9Q6Genspark's CEO says "just prompt it." My reaction? Microsoft Office is in trouble as more people discover the new way to work. They won't be going back.
-
Spherical Equivariant Graph Transformers for 3D Molecular Modeling
By
–
"A Complete Guide to Spherical Equivariant Graph Transformers" This paper explains how to make neural networks that understand 3D geometry the way physics does, so models for molecules and proteins give the same and correct answers no matter how you rotate the structure.
-
Managing Deepagents Memory: Learning and Reflection Over Time
By
–
Managing Deepagents Memory Memory allows agents to learn tasks, knowledge, or preferences over time. Here, we overview two ways to update Deepagent-CLI memory: direct instruction and reflection over past Deepagent sessions. : https://
youtu.be/3aS1A-0775s -
Meta Open-Sources PE-AV for Advanced Audio Separation Technology
By
–
We’re open-sourcing Perception Encoder Audiovisual (PE-AV), the technical engine that helps drive SAM Audio’s state-of-the-art audio separation. Built on our Perception Encoder model from earlier this year, PE-AV integrates audio with visual perception, achieving
-
Google AI Leaders Reflect on Year and Future Outlook
By
–
Join us live on X at 1PM PST / 4PM EST to hear @JeffDean
, @OriolVinyalsML
, @NoamShazeer
, and @OfficialLoganK reflect on our year in AI at Google, and explore what's coming next: -
Keras 3.13 Release: LiteRT Export, GPTQ Quantization, Adaptive Pooling
By
–
New Keras release: 3.13 Some major new features: • Model export to LiteRT (formerly TFLite) for mobile/edge
• GPTQ quantization support for post-training compression
• New Adaptive Pooling layers for dynamic architectures -
Introduction of FunctionGemma for Function-Calling Tasks
By
–
Not Gemma 4 but functiongemma-270m-it "FunctionGemma is intended to be fine-tuned for your specific function-calling task, including multi-turn use cases."