Two ways to get it: 1/ Register for TODAY's free AI conference (8,000+ joined). Once you book, you'll get the Gemini Mastery Guide as a bonus: https://
conf.cosprints.ai/?8 30 speakers from Google AI, AWS, Microsoft, Deliveroo. Starts 11am NYC. 2/ Grab the guide directly here:
MACHINE LEARNING
-
Free AI conference with Gemini Mastery Guide bonus
By
–
-
Token Undertrain and Variance in Language Model Training
By
–
Many papers! The tokens were significantly undertrained already last year, and the trend towards even fewer tokens likely just makes the variance worse! NanoGPT's small competition hides variance problems by gathering statistics over many seeds, IMHO just switch to medium size.
-
World Models in Robotics Training
By
–
True. World Models is what everyone in robotics is building. Trained on video data. See my latest post.
-
Raleigh Uses AI Vision and Digital Twins for Safer Streets
By
–
.@RaleighGov is proving how fast-growing cities can turn real-time AI into safer streets and smoother mobility.
— NVIDIA AI (@NVIDIAAI) 21 janvier 2026
By combining vision AI, digital twins, and AI agents, the city is transforming traffic management and urban response:
🚦 95% accurate vehicle detection to improve… pic.twitter.com/ZcdxjPYRwE.
@RaleighGov is proving how fast-growing cities can turn real-time AI into safer streets and smoother mobility. By combining vision AI, digital twins, and AI agents, the city is transforming traffic management and urban response: 95% accurate vehicle detection to improve -
New ‘Mini Companion’ Grok model appears, fast but old rollout mistake
By
–
A new “Mini Companion” model option appears on Grok for iOS. It feels like an older Grok model and likely a mistake rollout. Super fast though
-
The future of personal AI assistants is here, and it’s incredibly fun and addictive.
By
–
"Being able to make my computer do things – anything – by just talking to an agent running inside it is incredibly fun, addictive, and educational (…) we’ve barely begun to tap into the potential of personal assistants. There’s no going back." https://
macstories.net/stories/clawdb
ot-showed-me-what-the-future-of-personal-ai-assistants-looks-like/
… -
Why AI codes well in 2026 according to this post
By
–
Which AI have you used recently to believe that AI still codes like crap in 2026? Because there have been some serious advances and its vision is not far-fetched, I think.
-
Cold weather prompts GPU usage to fine-tune large XGBoost models
By
–
It's getting pretty cold over here. Time to fire up all the GPUs and fine tune some big XGBoost models. @trainxgb
-
LeCun: child perceives more visual data than LLM
By
–
This is false: LeCun does the calculation in a video, I need to find it: a child in their first 4 years of life perceives more visual data than the largest LLM trainings use.
-
Reflecting on the black-box nature of neural network weights
By
–
Yeah, I've been reading through the source code too. It's quite interesting. What's really interesting is that a lot of it is run by neural networks that we can't see inside of; we don't know the weights, for instance, but we can see the effect. My feed is generally better today