@MsftSecIntel SecOpsAl Supply Chain Alert in https://t.co/2YPOfjG7MS discovered this compromised package at 3:08 AM İstanbul time, everyone should have this free open source security tool.
mHC and AttnRes demonstrate that richer depth mixing improves model performance at scale. Is there a unifying perspective? 🤔
Meet our Depth Memory framework: each layer reads from a set of streams to form its input, then writes back its output.
Linear Depth Memory (LDM) equips models with a matrix-valued memory.
Work the latest evidence-backed detections first. Each row exposes severity, confidence, environment impact, evidence state, ownership, and the next safe action without turning scanner output into an automatic verdict. https://t.co/lbOuKLuAaz
A few years back some people from NSA and CyberCom released a paper & talk on "industrial-scale" vulnerability research. Imagine that now being augmented by a many-agent tech setup like this one used by Google.
Formidable.
https://t.co/54QEWtztRL
Sleep deprivation does more than reduce energy. It changes the brain you use to make decisions. When executive control deteriorates, impulses face less resistance and familiar patterns become easier to execute. Protecting sleep is therefore partly an act of behavioral control.
Can’t wait for the AI to get so good none of this business about detecting AI matters anymore
Silverware used to be handmade, but nobody complains about their dinner fork is stamped by a machine
The quality of ideas matter
The important thing is people can eat
2027 will be the year of AI threat detection. Detection relies on very cheap models, and that’s happening sooner than I expected with Deepseek Flash and GPT-5.6 Luna.
It will be amazing.
Nigerian Americans are some of the most educated people in the US.
64% of US children born from Nigerian immigrants have at least a Bachelor's degree, roughly double the US average.
Releasing the model weights and technical report of Kimi K3.
Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.
New model architecture: 2.5x the intelligence per unit of compute, not just more params.
Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.
Model weights: https://t.co/7m7eEg6Y0B
Tech report: https://t.co/yeu6cjpMCT
Tech blog: https://t.co/YTfiMSNM1f
Releasing the model weights and technical report of Kimi K3.
Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.
New model architecture: 2.5x the intelligence per unit of compute, not just more params.
Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.
Model weights: https://t.co/7m7eEg6Y0B
Tech report: https://t.co/yeu6cjpMCT
Tech blog: https://t.co/YTfiMSNM1f
Firmly rooted in the Word of Jesus, reading the Gospel every day, never stop dreaming of a land where there is peace and fraternity, nor stop looking at the world with wonder, nor stop facing the future with the light of hope. Do not let anything or anyone steal your hope.