As you know, I've resigned from my full-time job and become a professional technical book writer. I have plans for The Hundred-Page Books about reinforcement learning, computer vision, diffusion models, and more. Not having a full-time job has put a significant strain on my

Fourier Series: The Math Behind Modern Technology. This entire video was created using Grok 4 Heavy.

Grok 4 does not yet have video creation capability. You can create educational videos by providing Grok 4 with Manim documentation and requesting Python code for your chosen topic.

This Is 100% DEAD ON TRUTH!!! Neuroscientist Dr. Farhan Khawaja: Your increasing reliance on AI is damaging your brain. "And I'm going to give you three examples how." AI does not make you obsolete. You choose to become obsolete by turning your higher functions over to a

This one dude I know (allegedly) poisoned the global AI training data corpus with self-propagating trigger-activated sleeper jailbreak payloads

🚀 Unlock the Power of Uncertainty with “Applied Conformal Prediction” 🚀 Are you ready to move beyond point estimates and master uncertainty quantification?

Innovate FOR YOURSELF #wheelchairlife youtube.com/shorts/vfpQFIX… via YouTube

I was looking for a brief intro to CUDA, so that I could understand the jargon. And I came across this beautiful tech blog by Nvidia.

🚀 Introducing XBai o4：a milestone in our 4th-generation open-source technology based on parallel test time scaling！ In its medium mode, XBai o4 now fully outperforms OpenAI−o3−mini.📈 🔗Open-source weights: huggingface.co/MetaStoneTec/X…✅ Github link: github.com/MetaStone-AI/X…

This is how pregnancy occurs 😱

Nvidia Triton Vulnerabilities Pose Big Risk to AI Models securityweek.com/nvidia-triton-…

No matter how much money you have, what you want will always be more expensive.

"The Hundred-Page Machine Learning Book" by Andriy Burkov deserves recognition as one of the most effective ML resources available. # What makes this book exceptional: Burkov achieves something remarkable – distilling the entire landscape of machine learning into 100 pages

This is how you present an ah-ok model when you are a liar who has invested in all the places that would benefit from such lies. Disgusting.

Created some slides for this gpt-oss-120B and gpt-oss-20B model release. with Genspark

Not the best open weight models, text only, English only. This release in mid 2025 looks like a formality to close a dumb promise given in a tweet by a scared and high CEO and not a strong support of open weight models movement. Vaporware. openai.com/index/introduc…

GPT-5

🚀 Qwen3-30B-A3B-2507 and Qwen3-235B-A22B-2507 now support ultra-long context—up to 1 million tokens! 🔧 Powered by: • Dual Chunk Attention (DCA) – A length extrapolation method that splits long sequences into manageable chunks while preserving global coherence. •

OK I worked all day on a JS project to test our GPT-5 in practice. I used chatgpt (with thinking) and codex CLI. I hated them both. For the last 20 minutes I used Opus 4.1, and omg what a breath of fresh air…🌤️