←
Psyho2,892 posts
Psyho avatar

Psyho

@FakePsyho

Humanity's Last Programmer; Game Designer; Problem Solver; past: OpenAI (Dota), Pro Competitive Programmer, Poker

◷ Joined May 2012▧ 469 Photos and videos
Psyho avatar
Psyho@FakePsyho·16 Jul 2025

Humanity has prevailed (for now!) I'm completely exhausted. I figured, I had 10h of sleep in the last 3 days and I'm barely alive. I'll post more about the contest when I get some rest. (To be clear, those are provisional results, but my lead should be big enough)

Post attachment
Psyho avatar
Psyho@FakePsyho·Nov 15

that was the most smoothbrained stream by far I wasted like 1,5h on a single puzzle, because I forgot the red beam x transparent rock interaction and then I spent like 40 minutes on the overworld trying (and failing) to access any new areas I've finally unlocked the 2nd part of the game, so hopefully the future streams are going to be more fast-paced; I'll try to keep this Tue/Thu/Sun schedule going forward

Psyho avatar
Psyho@FakePsyho·Nov 14

WARNING: small You Can See Everything spoilers Last year, after beating AI in a programming contest, I received hundreds of DMs. Most of them were regular congrats or interview requests. Some of them were a bit more unusual. The one that confused me the most came from Elizabeth Holmes's account. At the time, the account was completely dormant (previous post was from 2015) and I honestly thought it had been either hacked or sold. The message said she was organizing some event in California and whether I'm interested in solving hard problems. It sounded very suspicious considering she was in prison. In the end, my sleep deprivation won over my curiosity, and I never replied. I completely forgot about the whole thing until I saw the twitter scene in the film. Turns out, the account wasn't hacked and I probably missed my one chance to join Theranos 2.0 🙃

Psyho avatar
Psyho@FakePsyho·Nov 12

this thing goes viral every other month, usually accompanied by another video showing human performance on the same task and a caption like "ants are smarter than humans" I've looked at the actual paper, so let me ruin this for you: - researchers performed over 200 trials with different groups of ants and humans; humans were roughly 5-20x more efficient (metric = combination of total movement and rotation) than ants; this includes experiments where humans had restricted communication - calling it "collective intelligence" is a bit of a stretch; researchers tried to simulate ants' movement with a simple algorithm that doesn't do any explicit communication between the ants and got very similar results; essentially ants are performing a slightly biased random walk that just happens to work quite well for this very specific instance of pathfinding with an irregularly shaped object - the only interesting part about this paper is that the groups with more ants performed better than smaller groups; this improvement is explained by the provided algorithm used in the simulation so no, ants aren't smarter than humans; it still looks cool though

The collective intelligence of ants solving the “piano movers” puzzle. 📽️: Dreyer et al., PNAS 2025

0:30
Psyho avatar
Psyho@FakePsyho·Nov 5

worth noting that 1.9x faster improvement in AI capabilities is not 1.9x faster improvement in AI capabilities of the best publicly available model

1. Benchmarks provide one piece of evidence for acceleration. Since Oct 2025, AECI scores for Anthropic’s models have grown at 1.9x [95%: 1.5–2.8] the earlier rate. Analysis choices affect this estimate, and it remains unclear how benchmark scores map to real-world capabilities.

Post attachment
Psyho avatar
Psyho@FakePsyho·Nov 3

to be fair, the main offender here is twitter's "algorithm" there's currently way too much incentive to QT anything that's going viral; you've interacted with a single viral tweet? enjoy those 200 quote tweets flooding your feed for the next 24 hours

Psyho avatar
Psyho@FakePsyho·Nov 3

I'm honestly surprised to see so much hate directed at the authors of the Weight-Space Programming blogpost for overstating the results in their headline, yet so little directed at the mindless AI influencers who blindly amplified it. Sadly, it's the same game. As long as people happily spread any bs that might go viral, researchers will have a very strong incentive to oversell their work.

Psyho avatar
Psyho@FakePsyho·Nov 2

meh, the more you dig into each experiment, the less magical it gets; it feels like a loose collection of cherry-picked demonstrations with a grand narrative built around them Exp 1-2 are tiny (~20K params) with unusually easy to interpret architecture; Exp 1 is an MLP, where nearly all params are used in the first layer, so you can easily tell what kind of objects the network is trying identify Exp 5 is about designing a game strategy (we already know that Astra is extremely good at this) and then converting it to a tiny (~20K params) MLP; entirely doable by a human, but of course would require a lot of effort Exp 3 & 8 are closer to automated mechinterp using python and established techniques, rather than direct weight interpretation (at least according to chat) to be clear, those are genuinely cool results, but I think people are reading too much into them I don't think any of this should meaningfully change anyone's intuition about what models are capable of

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…

Post attachment
Psyho avatar
Psyho@FakePsyho·Oct 22

staying at an airbnb took me 10 minutes to find the switch for the lamp above the bed this is some escape room-ass level design

Post attachmentPost attachmentPost attachmentPost attachment
Psyho avatar
Psyho@FakePsyho·Oct 18

when streaming Order of the Sinking Star, I try to explain my thought process whenever I find a solution interesting, here's one example notes: - that was near the end of my stream and I was running very low on energy, so you might want to watch this at 1.5-2x speed; sorry for broken English, this is all improvised - if you don't know the rules: the goal in this level is to remove all of the goblins, they kill each other then they're on adjacent tiles; we can move them around by teleporting them through the mirror; our hero is rogue and unlike in the classic sokoban, rogue pulls objects instead of pushing them

5:03
Psyho avatar
Psyho@FakePsyho·Oct 18

if you're wondering why suddenly I was motivated to finish an ambitious benchmark: I'm procrastinating doing other things in my life; turns out I'll do anything except for preparing for an interview

Psyho avatar
Psyho@FakePsyho·Oct 18

a small announcement: I'm almost done with a puzzlescript-based benchmark; you can think of it as ARC-AGI-3.5, maybe even ARC-AGI-4; I expect it to be the most reliable way of assessing agents' abstract reasoning skills in a game environments I went through over 300 games, handpicked the most interesting and diverse set of 40, fixed a ton of bugs, went through multiple iterations of prompt design, etc most games can be completed in around 1h, some of them are significantly harder; models are tested using standard harnesses (codex/claude code) with no code generation allowed; action limits are very generous and unfortunately the benchmark is already somewhat saturated, but the models are still insanely action inefficient FYI, I won't release the game set to the public, but I guess I might share runs from some of the rejected games. btw, if anyone knows, what do I have to do in order to not pay $10K out of my own pocket for testing, I'm all ears

Psyho avatar
Psyho@FakePsyho·Oct 17

Remek is currently working on a decision model (i.e. "system one" model, similar to Jev / OpenAI's Decisions API) and he's currently ranked #1 in open weights (#2 overall!) on JevBench, essentially beating almost every commercially released product, including Jev itself that went viral some time ago what's even crazier is that he's working on it alone, as an after-hours side project, he started a few (two?) weeks ago and he's paying for compute out of his own pocket

basal-omni 🇵🇱 - TOP1 🤩🤩🤩 🌍 na świecie w modelach otwartych oraz TOP2 wśród zamkniętych i otwartych Wg niezależnego benchmarku modeli system-one gdzie testowane są wszystkjie modele na świecie - JevBench (link do bencha w pierwszym wpisie). Czyli można? Można! Polskie AI 🇵🇱 basal wyprzedził model OpenAI, Microsoft oraz pierwowzór z TypeSafe (jev). Thank you, @airesearch12 , for the independent evaluation on JevBench.

Post attachment
Psyho avatar
Psyho@FakePsyho·Oct 15

I suspect that one of the biggest reasons why almost no one pursued LLMs despite GPT-2's overnight success was surprisingly simple: almost no one was excited to work on language even at OpenAI, people didn't want to leave their cool game environments behind

Psyho avatar
Psyho@FakePsyho·Oct 15

a few random facts from the development: - the bot was still not good enough to beat the pros a few days before the showmatch; we were still making last-minute changes during The International; in particular, I was making a ton of changes to feature extraction, while Jakub & Szymon were surgeoing old weights to the new model - we didn't have time to test our changes, so we literally pushed everything directly into "production" because we only had enough time for one training run; it worked - we were aware of multiple exploits you could do vs the bot (our training env wasn't general enough and you could do a lot of weird things like luring the bot to other towers); that being said, I was sure that it's nearly-impossible to win vs the bot in a straight up micro battle - as long as Dendi tried to play vanilla 1v1, I thought we have 98%+ chance of winning - a lot of pros thought that we have "solved" the items that heroes should buy in 1v1; in particular the bot was buying a ton of consumables; AFAIK pros were later inspired by the bot and the meta shifted to buying more consumables in the midlane; the truth is, we just hardcoded the item order and never had enough time to explore other options