NotYerSheeple🏴‍☠️|LLM-Whisperer|Powered by Pho

Unpopular Opinion: The only dumb questions are the ones you didn't ask because you were afraid of others calling you 'dumb.'

@The_Real_Fly it ain’t gonna spend itself…

US is spending at least $2b per day on Iran war -- FOR WHAT?

@johncrickett @marzeaned finite resources are a bitch.

@marzeaned CLAUDE.md isn't rules, it's context. People spending ages planning swarms of subagents with different skills productivity is theatre.

I spent the weekend actually reading the Claude Code docs. It's a rabbit hole. CLAUDE.md files. MCP configs. Skills. Subagents. Hooks. Plugins. Agent Teams. You could spend more time configuring Claude Code than building software. All of it is productivity theatre. The only thing that actually matters: think first, then give it focused, relevant context.

@CynicalPublius @E0_DS0_Omega Everyone has a plan. Until they get punched in the face. What do they call the team tasked with coming up with shit no one thought of?

@E0_DS0_Omega Agree 100%. But to say we do not engage in incredibly detailed planning is untrue. And the process of that planning is a big part of how we react when the initial plan goes awry.

The Department of War is populated by literally tens of thousands of mid-grade officers whose singular job in life is to write and continuously update a plethora of operations orders that cover every imaginable contingency involving the US military, down to the tiniest detail. If you can imagine it, there is an OPORD for it. Never tell me a US military mission lacked adequate planning. It's a lie.

You're a 15 second sliding context window with the working memory of a goldfish, your long term memory is mainly fake, and it's a minor miracle you can get out of the door in the morning.

@anduflas @TheStudyofWar if Iran ever got a nuke you’d understand. this is long overdue. anyone within range of Iranian missiles, and a brain, is secretly cheering on Israel and the US. However ‘bad’ it is now it would be worse later. oil market is global and if they nuked the Saudi oilfields world fuckd

@TheStudyofWar As an American taxpayer, I don’t understand why we are fighting Isreal’s war when American kids go without free school lunches and adequate healthcare. Let Isreal defend themselves.

NEW | Special Report: The war in Iran is currently in a phase in which the military trajectory is relatively positive: the United States is steadily destroying Iran’s ability to use its most essential tool in the war — drone and missile attacks — which in turn underpin the entire Iranian strategy. Iran has still done some damage to US forces, and it is still firing drones and missiles, though the overall attack rate is slowly decreasing. These attacks still pale in comparison to the major attacks Iran sought to conduct in an existential war and have caused neither operationally significant damage nor widespread casualties. The US-Israeli combined force will need time to achieve its military objectives and prevent Iran from inflicting further political and economic pain upon the United States and its allies in the region, but the campaign remains incomplete, and it is too soon to forecast its outcome. Declaring it an operational failure is unquestionably premature.

@joeroganhq it’s incorrect

Elon Musk: "If the simulation theory is true, then it is very likely that only the most interesting outcome is the most likely. Only the simulations that are interesting will continue, the simulators will stop any simulations that are boring."

@VentureCoinist @himgajria UBD - Universal Bug Diet

@himgajria also possible we get a much more dystopian version like UBT (universal basic tokens) and as ai both physical & virtual do more of our jobs the only way to earn is by renting/using them with your allocated tokens many rely on the ubt handout...gap grows 1000x more than covid

Prediction: We won’t have UBI. We’ll have UCI. (universal credit income, or whatever else it might be called) Assuming labour will cease to exist, governments will have to pay you money, for you to fulfill the consumption side of the economic equation. While everyone believes that people will get an equal paycheck similar to stimulus payments, I believe it’ll be akin to airdrops, where each person is credited based on their behavior, actions, social score, etc. TL;DR: take the chinese social credit system and combine it with UBI.

@tomfgoodwin after trying instacart, i was not happy constantly fielding texts about substitutions. gave up on that kind of service. just doesn’t work for me.

What happened to ultra fast grocery delivery? did they all go bust silently? I've not seen GoPuff etc for ages,.

@FinanceLancelot great way to encourage communication and exchange of ideas - shut them down 🤡

@IngrahamAngle you are years too late

“This sports BETTING trend has become an epidemic.” Eli Thompson warns many students are losing serious money. “I’ve seen friends lose $400 or $500 — not just their own money, but money set aside for college or a car. And it’s affecting school, sports, and their social lives.”

@StephenPunwasi it’s called global geopolitics. once you see it, you can’t unsee it. it is what it is. it’s been going on forever. it will keep going until the asteroid.

It’s true. Murdering people and taking their stuff is quite profitable. aljazeera.com/amp/news/2026/…

@shen_shiwei i heard we’re renaming it the Strait of America.

Because the Strait of Hormuz is not U.S. private property!

@mattvanswol why do you think they fought him so hard, tried to lock him up or kill him…?

I cannot get over the fact Trump just stopped illegal immigration. By himself. wtf.

@DeItaone @JCOviedo6 The Turing Police have arrived!

WHITE HOUSE READIES EXECUTIVE ORDER TO WEED OUT ANTHROPIC - AXIOS

@ElectionWiz this is not the correct question. the correct question is who’s next?

President Trump says the Iran war will be over very soon. What do you think that means?

@elonmusk @wholemars @karpathy @farzyness there is no singularity. infinity is not reachable.

@karpathy @farzyness We are in the Singularity

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.

@sweatystartup i think you’ll need to audit the incomes of the authorities/politicians in charge and you’ll find your answer

It’s the drugs. Take the drugs out of california and other major cities and crime and homelessness drops 99%. How hard is it in a modern society to cut off and punish drugs vs enabling the use?

@FinanceLancelot the official stats disconnected from the real economy a long time ago. we never recovered from the 2008 aborted liquidation. i get the sense they think they outlawed it. true, until it’s not anymore. no idea when that is.

There's a plethora of economic data coming out this week that will be cooked. This is probably the narrative they'll use to pump markets and say the economy is "booming." Inflation data, GDP growth, unemployment, job openings, etc. You can smell the burning smell from here.

@tomfgoodwin copycats ;) my favorite trick was the VirtualBox trojan horse where the core framework was free, but once you installed the toolkit extensions - bam! they got ya. I gotta respect the art.

@NotYerSheeple And Broadcom

People talking about the SAAS apocalypse don't get this is how it actually happens. - Hi, we can vibecode your product in 2 months, so how about you cut your annual fees down from $500k to $200k for us. So we don't have to make it. - Maybe, but how about we do a license audit, where will sue you for $4m for breaking T&C's in 9 global offices. BTW, have you seen how much time we spent on support, and training, and we're embedded so deeply into every system, it will take you years to get rid of us. - OK, 500K it is, and see your at the F1, we loved your box last year. - It's going to be $600k this year. -OK, but we still get your box tickets? right?

@FirstSquawk @JCOviedo6 starve the beast

ANTHROPIC CFO: U.S. GOVERNMENT ACTIONS COULD CUT 2026 REVENUE BY MULTIPLE BILLIONS ACROSS THE BUSINESS — COURT FILING.