Taking a short break … but will soon be back to our deep thinking mode! Thanks to @demishassabis for featuring some of our works from IMO 2024 (#AlphaGeometry, #AlphaProof) to #DeepThink + other amazing works from colleagues at @GoogleDeepMind
!
MACHINE LEARNING
-
DeepMind Returns to Deep Thinking Mode with Advanced AI Research
By
–
-
Google DeepMind Launches Deep Thinking Model at I/O
By
–
It's been a wild run over the last 2 weeks to land Deep Thinking just in time for I/O! Heroic efforts across many teams @GoogleDeepMind from research to production that felt almost impossible last month! Proud of our research team for the relentless effort over the past 6 months
-
Google introduces Gemini 2.5 Pro Deep Think model outperforms Pro
By
–
BREAKING : Google introduced Gemini 2.5 Pro Deep Think – a new model that outperforms the normal Pro version. It will become available to trusted testers on GCP Vertex via an API.
-
Translating Cutting-Edge Research Into Real Products
By
–
A consistent theme we’ll talk about today is the idea of taking cutting-edge research and making it a reality. Here are a few examples of research efforts making their way into new and existing products
-
ARC-AGI-2 Paper Published: Design, Challenges, Human and Model Performance
By
–
We just posted our paper on ARC-AGI-2 — covering its design principles, what makes it more challenging, a detailed analysis of human performance, and current model performance.
-
Adjoint Sampling: New Scalable Algorithm for Reward-Based Generative Models
By
–
Introducing Adjoint Sampling, a new learning algorithm that trains generative models based on scalar rewards.
— AI at Meta (@AIatMeta) 20 mai 2025
Based on theoretical foundations developed by FAIR, Adjoint Sampling leads to a highly scalable practical algorithm, and can become the foundation for further research… pic.twitter.com/RyVVzbEfBjIntroducing Adjoint Sampling, a new learning algorithm that trains generative models based on scalar rewards.
Based on theoretical foundations developed by FAIR, Adjoint Sampling leads to a highly scalable practical algorithm, and can become the foundation for further research -
Visual Planning: Thinking Only with Images Research
By
–
Visual Planning: Let's Think Only with Images Xu et al.: https://
arxiv.org/abs/2505.11409 #ArtificialIntelligence #DeepLearning #MachineLearning -
Keras Release: Model Sharding, Muon Optimizers, New Ops
By
–
New Keras release is out — includes saved model file sharding for very large models, the Muon optimizers, new ops and many performance improvements
-
Voyager SDK Enables Edge AI Computer Vision Deployment
By
–
Build #AI at the edge without the complexity.
The Voyager SDK from @AxeleraAI makes it easy to deploy high-performance computer vision apps using our #modelzoo or your own models. New video: Quick intro, demos, real-time inference & more – https://
eu1.hubs.ly/H0kfpHm0 -
Reinforcement Learning Overview by Kevin Murphy
By
–
Reinforcement Learning: An Overview Kevin Murphy: https://
arxiv.org/abs/2412.05265 #ArtificialIntelligence #DeepLearning #MachineLearning