#LLM
Curated coverage on «LLM»: releases, research, benchmarks, open-source.

Omni-Woodpecker Cuts Hallucinations in Omni-LLMs Without Retraining: +12.9 on CMM
A Nanjing University team proposes OWP, a method that corrects an omni model's hidden states before decoding and adds 12.9 points on CMM for Qwen2.5-Omni-7B.

Researchers Explain How JEPA Selects Features: Relationships Matter, Not Lone Features
Mosh Levy, Aly Lidayan and Yoav Goldberg introduce "predictive closure": JEPA retains not individually predictable features, but sets of features that predict each…

Four Chinese Players Are Building Real-Time Interactive Video Worlds
Alibaba's HappyOyster, Ant's LingBot, Vidu S2 and PixVerse R2 put interactive video worlds at the center of China's AI market: the model reacts to user actions as…

Attempt on Sam Altman's Life: An Ideologically Motivated Arson Over Fear of AI

Three AI Inflection Points: From ChatGPT to Agents and Physical Intelligence

Google Expands Gemini's Personal Intelligence to India

Claude Mythos Is Big Tech-Only, CaveMan Cuts Token Spend Threefold

How to Turn Your Best AI Prompts into Quick Tools for Chrome
The Skills extension lets Chrome users save, edit and run complex AI prompts with one click, turning the browser into a powerful automation platform.

Microsoft Takes Over OpenAI's Stargate Data Center in Norway
Microsoft has agreed to lease capacity at a Norwegian data center originally built for OpenAI's ambitious Stargate AI project.

AWS's Path-to-Value Framework: Getting AI Into Production

Amazon Launches Optimized AI Model Deployment in SageMaker JumpStart

Amazon SageMaker HyperPod: Optimizing AI Model Inference and Cutting Costs

Google Scales Up "Personal Intelligence": Gemini Moves Deeper Into Users' Lives

Researchers Hack AI Agents on GitHub to Steal Credentials
Researchers found critical flaws in AI agents running on GitHub. A new class of prompt-based attacks can quietly steal user credentials, while the tech giants stay…

Anthropic to Open Its Mythos AI Model to UK Banks
Mythos, an AI model specialized in finding cyber threats, is moving into the UK financial market after an exclusive launch in the US and on Wall Street.









