Frameworks come and go. XGBoost is forever. @trainxgb
MACHINE LEARNING
-
Robotics and AI at Scale: World Reconstruction and Neural Networks
By
–
Drops plenty of clues about the evolving "robotics + AI at scale" framework, featuring world reconstruction tools, simulators that "dream up" physical interactions, reinforcement learning & core neural-net-driven elements. A car is essentially just another robot. Hence S & X deprecation.
→ View original post on X — @reza_zadeh, 2026-02-03 04:18 UTC
-
Exploring Multi-Agent Ecology and Shared Knowledge Scratchpads
By
–
Absolute bars: “Sites like moltbook function as a giant, shared, read/write scratchpad for an ecology of AI agents – how might these agents begin to use this scratchpad to a) influence future ‘blank slate’ agents arriving at it the first time, and b) unlock large-scale
-
SAS Predictions 2026: 15 Expert Insights on Data and AI
By
–
Time is of the essence. Now is a good time to see what could be coming down the pipeline this year for data and AI professions. Here are 15 predictions at the ready from experts across SAS: http://
2.sas.com/6013CF9DU #SASPredictions -
Tesla FSD: Deep Dive into AI with Ashok Elluswamy
By
–
Tesla FSD deep dive, by VP of AI at Tesla, Ashok Elluswamy @aelluswamy
— Reza Zadeh (@Reza_Zadeh) 2 février 2026
One big neural network consumes video, sound, telemetry, basic maps, & outputs at 36hz steering & speed. All run on an custom chip on car. Outstanding fleet data management, and way more! cc @elonmusk pic.twitter.com/rvboZRl7FwTesla FSD deep dive, by VP of AI at Tesla, Ashok Elluswamy @aelluswamy
One big neural network consumes video, sound, telemetry, basic maps, & outputs at 36hz steering & speed. All run on a custom chip on car. Outstanding fleet data management, and way more! cc @elonmusk [Translated from EN to English]→ View original post on X — @reza_zadeh, 2026-02-02 22:24 UTC
-
Codex incorrectly suggests neural network for classical ML algorithm
By
–
I've been trying to reimplement a very simple *classical* ML algorithm (no, not XGBoost) in a different programming language. I've been using Codex for it. It keeps resorting to writing a *neural network* implementation, even though that approach is several orders of magnitude
-
Real FSD downstream from AGI, any modern car could be
By
–
I’ve been saying this for years. *REAL* FSD will just be downstream from AGI. As computational power and models improve, almost any modern car could *in principle* be turned into FSD.
-
A Quick Primer on World Models and Hybrid Approaches
By
–
quick 20 min primer on world models incld. the hybrid approach (also on YT) https://t.co/7wc6NvUHOF
— Bilawal Sidhu (@bilawalsidhu) 2 février 2026quick 20 min primer on world models incld. the hybrid approach (also on YT)
-
Kaggle and Google DeepMind Introduce New Benchmarks for AI Agentic Capabilities
By
–
In partnership with @GoogleDeepMind
, @Kaggle announced today an expansion of the #Kaggle Game Arena, introducing new benchmarks designed to test critical #AI skills required for real-world agentic capabilities. Key updates include:
New werewolf benchmark
New poker benchmark -
Gemini 3 Models Lead Chess Leaderboard with Rapid Improvement
By
–
Also cool to see our latest Gemini 3 models at the top of the chess leaderboard – incredibly rapid improvement from when the benchmark first launched! Note all models still only play at weak amateur level so lots of improvement needed.