31.01.2026
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) 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.
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.
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.com/newsroom/2026/01/apple-reports-first-quarter-results/
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?)
“There isn’t a lot of reliable information out there about how to buy a gas mask, especially for the specific purpose of living under state repression. But hopefully after reading this guide you’ll feel equipped to make an educated decision.”
“AI skeptics need to update their priors: Plenty of cause for concern, plenty of room to hit these companies for unethical behavior, resource demands, etc, but we are so, so far past the era of ‘stochastic parrots’”.