NVIDIA releases open-source robot world model trained on 44,000 hours of human video https://perplexity.ai/page/nvidia-releases-open-source-ro-O3KP1lsGStOa5.u4nkDGyw @NVIDIA
#AI #MWC26
RESEARCH
-
NVIDIA Releases Open-Source Robot Model Trained on Videos
By
–
-
Unified Latents: DeepMind’s Automatic Latent Information Training
By
–
now trending on alphaXiv "Unified Latents (UL): How to train your latents" in latent diffusion you’re choosing how much information the latent should carry, and there’s no automatic “right” setting. But in this DeepMind paper, they showed that you don't have to guess
-
SwingArena: AI Models Evaluated on Real Software Engineering Tasks
By
–
Can AI models handle the complexity of real-world software engineering beyond simple code snippets? Researchers from HKU, UCLA, LMSYS Org, and several global institutions just released SwingArena. It is a competitive evaluation framework that mimics a professional development
-
Human neurons fire similarly to synthetic neural network neurons
By
–
you know human neurons fire the same way neural networks synthetic neurons fire right?
-
Neural net learns like humans, comments miss the point
By
–
yes read the comments, all of a sudden it’s fine. they don’t understand a neural net learns similar to how we learn, therefor they are not keeping up
-
Meta’s PAHF Framework Personalizes AI Agents Through Human Feedback
By
–
New research from Meta. AI agents are powerful but generic. Personalized agents are on the rise, but hard to get right. This new research introduces PAHF (Personalized Agents from Human Feedback), a framework that teaches agents to learn and adapt to individual user
-
Claude Sonnet 4.6 and Gemini 3.1 Pro Launch This Week
By
–
AI Explosion This Week: Models Upgraded, Tools Launched, Drama Unfolds! – Claude Sonnet 4.6 from @AnthropicAI drops as the new default for free/Pro users—beats Opus 4.5 in prefs, crushes coding & agent tasks with top safety scores.
– @Google unleashes Gemini 3.1 Pro: Tops -
GRAIL Cancer Detection Trial Fails to Meet Primary Endpoint
By
–
On the huge randomized GRAIL trial
Front page @nytimes today @RebeccaDRobbins "The study’s goal was to show a 20 percent reduction in advanced cancers among those who got the test. That did not happen."
gift link https://
nytimes.com/2026/02/20/hea
lth/cancer-detection-test-grail.html?unlocked_article_code=1.N1A.Ivyd.lMF48TnsCUBK&smid=url-share
… -
Graph Machine Learning 2nd Edition with PyTorch Geometric Techniques
By
–
2nd Edition — now at http://
amzn.to/45Y3LyI v/ @PacktDataML Graph Machine Learning — Latest advancements in Graph Data to build robust #MachineLearning algorithms 𝓚𝓮𝔂 𝓕𝓮𝓪𝓽𝓾𝓻𝓮𝓼:
Master new graph ML techniques through updated examples using PyTorch Geometric and