When horrible people make bad art, the reviews are fun to read. Melania doc review: “Trump film is a gilded trash remake of The Zone of Interest” and “it’s one of those rare, unicorn films that doesn’t have a single redeeming quality”.
“The end of Temporary Protected Status has...
“The end of Temporary Protected Status has Holocaust survivors offering to hide Haitian staffers, according to the CEO of a senior-living center in Florida.” The CEO of the center: “That reminds me of Anne Frank.”
Illustrator Chris Piascik has made several of his...
Stamps: Graphic Art in Miniature
The Global Story: The mystery of Putin and the apartment bombs
31.01.2026
Britain's former prince Andrew faces fresh allegations
#490 – State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI
Quoting Andrej Karpathy
Originally in 2019, GPT-2 was trained by OpenAI on 32 TPU v3 chips for 168 hours (7 days), with $8/hour/TPUv3 back then, for a total cost of approx. $43K. It achieves 0.256525 CORE score, which is an ensemble metric introduced in the DCLM paper over 22 evaluations like ARC/MMLU/etc.
As of the last few improvements merged into nanochat (many of them originating in modded-nanogpt repo), I can now reach a higher CORE score in 3.04 hours (~$73) on a single 8XH100 node. This is a 600X cost reduction over 7 years, i.e. the cost to train GPT-2 is falling approximately 2.5X every year.
Tags: andrej-karpathy, gpt-2, generative-ai, ai, llms, openai
I learned two important pieces of news in this post: 1)...
I learned two important pieces of news in this post: 1) Frozen OJ from concentrate as a product is being discontinued by major producers. 2) Beverage analysts refer to market share as “share of throat.” And I think that’s just lovely.
The Happy Pod: What makes people instinctively kind?
Next week, Sa. 07.02.2026, I’ll be playing my first gig with Brutal Unrest in Aachen. If you want to know what this might sound like, the last album is on Bandcamp.
Singing the gospel of collective efficacy
Quoting Steve Yegge
Getting agents using Beads requires much less prompting, because Beads now has 4 months of “Desire Paths” design, which I’ve talked about before. Beads has evolved a very complex command-line interface, with 100+ subcommands, each with many sub-subcommands, aliases, alternate syntaxes, and other affordances.
The complicated Beads CLI isn’t for humans; it’s for agents. What I did was make their hallucinations real, over and over, by implementing whatever I saw the agents trying to do with Beads, until nearly every guess by an agent is now correct.
— Steve Yegge, Software Survival 3.0
Tags: steve-yegge, coding-agents, generative-ai, ai-agents, ai, llms
Apple Reports Record-Breaking Revenue and Profit for Q1 FY26
apple.com/newsroom/2026/01/apple-reports-first-quarter-results/
Proof Without Content
Moltbook is the most interesting place on the internet right now
Iran says it's ready for talks with US
This guy built an autonomous flying umbrella (powered...
This guy built an autonomous flying umbrella (powered by drones) that automagically follows you around in the rain. (A possible counter to a personal raincloud?)