👾John Lagerling

@ky__zo time to create AI CFO

@pmarca It’s a temporary illusion indeed

TLDR There is no inner self, you're chasing an imaginary concept, the end.

Had a parent-teacher conference this morning My wife told me not to come I came anyway She said "please just listen and nod" I said "I always listen" She said "you listen like you're sitting in a boardroom looking for something to challenge" That's how listening works Nice classroom Small chairs I am 6'4" and was seated at a desk designed for someone who still believes in Santa Claus My knees touched my chest The teacher introduced herself Shared her identified pronouns I shared my identified adjectives Smart and handsome My wife closed her eyes The teacher had a folder Color-coded tabs I respected the organization She said our son is "a pleasure to have in class" My wife smiled I waited That sentence is never the whole report It's the executive summary before the risk section She said "however" There it is She said he "asks a lot of questions" I said "good" She said "during quiet time" I said "when is quiet time?" She said "it's when students are expected to work independently and in silence" I said "so he's the only one trying to get information and you've structured the environment to prevent it?" My wife put her hand on my arm I continued The teacher said he recently told another student that "sharing pencils doesn't make sense if nobody brings their own" I said "that's an accurate observation" My wife squeezed harder The teacher said she's concerned about his "resistance to group activities" I said "he's not resistant. He just doesn't see the value of doing more work for the same grade." The teacher said he also corrected her math on the whiteboard I said "was he right?" She paused She said "that's not the point" I said "it's a little bit the point" My wife stood up Sat back down Compromise The teacher pulled out an evaluation sheet Categories like "works well with others" and "follows directions" and "respects classroom norms" All subjective Not a number on the page I asked how these are graded She said "based on observation" I said "so one person's opinion with no second review?" She said "it's professional judgment" I said "my auditors say that too. Right before I disagree with them." She looked at my wife My wife said "I'm sorry about him" I said "I'm sitting right here" My wife said "I know" The teacher said overall he's a bright kid and she just wants to make sure he learns to "collaborate" I said "collaboration is important. But so is recognizing when you're the only one doing the work. He'll learn that again in college. And again in the real world. Might as well start now." Nobody spoke The teacher closed her folder She said "I think we've covered everything" I said "one more thing" She braced herself I said "his reading is above grade level. His math is strong. He asks hard questions and corrects mistakes when he sees them. I just want to make sure this school knows what it has." The teacher looked at me differently My wife looked at me differently I said "that's all" We left In the car my wife was quiet Then she said "he's turning into you" I said "is that a good thing?" She didn't answer From the backseat he said "dad, why does the teacher count off for asking questions? Isn't that the whole point of school?" I looked at my wife She looked out the window I said "yes. It is." He said "I don't think she likes when I'm right" I didn't say anything Neither did my wife Small chairs Color-coded tabs No follow-up items But the kid's going to be fine Sent from my iPhone

@Bardissimo She needs braces. And an ego-check.

Greta Thunberg is absolutely bonkers mad. A narcissistic sociopath if ever there was one. God knows how much the actual peoples of Iran, Palestine and Cuba will hate this Hamas- mullah- och and communist-hugging bitch when this is all over.

Men socialize by insulting each other, but they don’t really mean it. Women socialize by complimenting each other, and they don’t mean it either. 😂

@BlueNoteTokyo Great performance!

《 Welcome back to Japan🇯🇵 》 ロバート・グラスパーが日本に到着しました！ ブルーノート東京公演は明日よりスタート。現代ブラック・ミュージック・シーンの重要人物が魅せるプレミアムな5日間をお楽しみに！ ROBERT GLASPER - PIANO TRIO 3.12 thu., 3.13 fri., 3.14 sat., 3.15 sun., 3.16 mon. bluenote.co.jp/jp/artists/rob… #robertglasper #bluenotetokyko @robertglasper

Grok has finished reviewing the entire corpus of active EU legislation. It recommends deleting 89%. @_FriedrichMerz

You are not your grand plans. You are your daily patterns.

The token cost to build a production feature is now lower than the meeting cost to discuss building that feature. Let me rephrase. It is literally cheaper to build the thing and see if it works than to have a 30 minute planning meeting about whether you should build it. It’s wild when you think about it. This completely inverts how you should run a software organization. The planning layer becomes the bottleneck because the building layer is essentially free. The cost of code has dropped to essentially 0. The rational response is to eliminate planning for anything that can be tested empirically. Don’t debate whether a feature will work. Just build it in 2 hours, measure it with a group of customers, and then decide to kill or keep it. I saw a startup operating this way and their build velocity is up 20x. Decision quality is up because every decision is informed by a real prototype, not a slide deck and an expensive meeting. We went from “move fast and break things” to “move fast and build everything.” The planning industrial complex is dead. Thank god.

It takes ten years, minimum, to move into a house.

@OlofFlodin Att lägga ned är ett bättre alternativ.

@bryan_johnson Terrence McKenna - just channel him

I'll be live streaming the 5-MeO-DMT. Am putting together the line up. Who should we have call in?

@OliverMolander 100% accurate

As a Finnish-Swede who as lived over a decade in Stockholm, I can confirm that this one is pretty correct.

“why do you have a gap in your resume?” idk why is there a gap in your staff? something to worry about

@forallcurious Shoot I just extracted my last one LITERALLY YESTERDAY

🚨: Wisdom teeth contains stem cells that aid in repairing the heart, brain and bones, recent research says.

@Yuchenj_UW @karpathy Feels like the AlphaGo moment for LLM pokegym trainers

Having fun with @karpathy’s autoresearch. I told Claude Code: “You’re the chief scientist of an AI lab with 8 GPUs. You’re Andrej Karpathy. Run parallel experiments and decide what to try next.” It edited program.md, ran for 11+ hours, and completed 568 experiments. Each experiment uses 1 GPU. Every round the “chief scientist” reviews the previous round of 8 results and designs the next 8 experiments. It's interesting to see the Claude agent, the chief scientist evolved a 3-phase strategy: Phase 1. Broad Exploration Early rounds explore many axes: architecture, optimizer, LRs, ablations. Phase 2. Focused Refinement After easy wins dry up, it runs deeper sweeps (e.g. 5 GPUs sweeping RoPE base 30k → 500k in one round). Phase 3. Heavy Validation Later, 50–75% of GPU budget goes to seed variance checks instead of new ideas. I feel it's overkill tbh. I'll keep the chief scientist running to see if it transfers to larger models and beats Andrej’s new "Time to GPT-2" leaderboard winner.

AI Psychosis or not, this is art.

Three days ago I left autoresearch tuning nanochat for ~2 days on depth=12 model. It found ~20 changes that improved the validation loss. I tested these changes yesterday and all of them were additive and transferred to larger (depth=24) models. Stacking up all of these changes, today I measured that the leaderboard's "Time to GPT-2" drops from 2.02 hours to 1.80 hours (~11% improvement), this will be the new leaderboard entry. So yes, these are real improvements and they make an actual difference. I am mildly surprised that my very first naive attempt already worked this well on top of what I thought was already a fairly manually well-tuned project. This is a first for me because I am very used to doing the iterative optimization of neural network training manually. You come up with ideas, you implement them, you check if they work (better validation loss), you come up with new ideas based on that, you read some papers for inspiration, etc etc. This is the bread and butter of what I do daily for 2 decades. Seeing the agent do this entire workflow end-to-end and all by itself as it worked through approx. 700 changes autonomously is wild. It really looked at the sequence of results of experiments and used that to plan the next ones. It's not novel, ground-breaking "research" (yet), but all the adjustments are "real", I didn't find them manually previously, and they stack up and actually improved nanochat. Among the bigger things e.g.: - It noticed an oversight that my parameterless QKnorm didn't have a scaler multiplier attached, so my attention was too diffuse. The agent found multipliers to sharpen it, pointing to future work. - It found that the Value Embeddings really like regularization and I wasn't applying any (oops). - It found that my banded attention was too conservative (i forgot to tune it). - It found that AdamW betas were all messed up. - It tuned the weight decay schedule. - It tuned the network initialization. This is on top of all the tuning I've already done over a good amount of time. The exact commit is here, from this "round 1" of autoresearch. I am going to kick off "round 2", and in parallel I am looking at how multiple agents can collaborate to unlock parallelism. github.com/karpathy/nanoc… All LLM frontier labs will do this. It's the final boss battle. It's a lot more complex at scale of course - you don't just have a single train. py file to tune. But doing it is "just engineering" and it's going to work. You spin up a swarm of agents, you have them collaborate to tune smaller models, you promote the most promising ideas to increasingly larger scales, and humans (optionally) contribute on the edges. And more generally, *any* metric you care about that is reasonably efficient to evaluate (or that has more efficient proxy metrics such as training a smaller network) can be autoresearched by an agent swarm. It's worth thinking about whether your problem falls into this bucket too.

@elonmusk It’s ok as it’s all a cosmic joke. A meaningful joke.

💯😂

The Science Fiction Decade.