real ugly/rich ₿ars

yes Im ugly, but I relax in the pool with the $MILF lol

@gubardy_DEFl Forcing binary outputs where uncertainty exists is probably root of lot of hallucinations

as developer, one of hardest parts of building with ai is that llm can suddenly become useless and keep repeating same mistakes in loop. at that point, you have to keep pushing it to try different direction and that is not only problem. runtime errors and random crashes in middle of session make it obvious how fragile whole system still is in real use, models often get stuck repeating same broken fixes over and over even after many correction prompts. they can hallucinate heavily, inventing functions libraries apis or syntax that do not exist. they can lose track of long conversations and forget requirements they had already confirmed. they can also produce inconsistent code from one iteration to next or suddenly stop with empty generic or irrelevant replies what is needed is real reset in how inference engines are designed qubics ternary model uses 1 for inhibitory, 0 for neutral or unknown, and 1 for excitatory, which lets it handle uncertainty and ambiguity directly instead of forcing everything into a binary choice like traditional systems because of that, root cause of hallucinations gets removed. when system does not have enough information, it does not need to make something up. it can stay in an unknown state and keep adapting continuous time processing, together with lower information loss during recurrent updates, also helps avoid infinite loops and sudden crashes, making inference far more stable and resilient in way that is closer to biological systems that do not break under pressure qubic is only ternary based infrastructure built to solve these hallucination problems #qubic

you are 21 and spend 3 hours day scrolling, and it seems harmless by 27 you cannot get through an article without reaching for your phone every couple of paragraphs by 33 you start wondering why nothing you begin ever gets finished, and that project you kept imagining is still only an idea. there is still never enough time by 40 you have not finished book in ten years and somehow your life already moved past you

@hehebox3r 💯

@work_bars_work The real unlock is AI connecting adjacent fields in ways humans overlooked. That kind of leap is fascinating.

23 y/o dude solved an erdos problem in chatgpt that mathematicians have been grinding on for 60 years, with just one prompt model pulled formula from an adjacent field that nobody had tried for this problem ai cyborgs, we are so back

hey everyone im cooked! > finally i got first X payout > the value is not important to me > my gratitude is I'm getting paid after 8x below minimum earnings. Now I can say that I officially work for Elon Musk 😭 If you want to know, what I do?🧵

@gubardy_DEFl yo... this is every long term holder nightmare

bought eth in 2019 because you believed in the vision held it for years thinking eth was headed to 50k moved it to a wallet, wrote the seed phrase on paper, and forgot about it 7 years later you open the wallet just to take a look balance is 0 eth you never signed anything you never clicked anything you were not even online 571 other people woke up to the same thing nobody knows how it happened nobody knows who did it nobody knows why the funds are already in monero

@stoner_races this is not future tense anymore....

some guy in china is reportedly making more than 30000 month with claude code basically acting like an employee claude code is running dozens of instagram accounts around clock while constantly researching hundreds of other accounts that are taking off honestly ai already looks like machine for printing money he pushes all the repetitive boring work onto ai and ends up being free to relax most of the day crazy times honestly ai driven automated monetization is already here

@mr_golf7 11 hour flight☹️😥

this chinese developer ran llama 70b locally on macbook during flight and spent full 11 hours working on client projects without internet he was sitting by window on transatlantic flight with macbook pro m4 and 64 gb of memory. onboard wifi cost 25 dollars for trip, and he passed on it no cloud api no connection to anthropic or openai no internet at all just local llama 3.3 70b in bf16 and a custom orchestrator script model was running through llama.cpp at 71 tokens per second with a context window of about 60000 tokens. memory usage sat at 48.6 gib out of 64, and battery at takeoff showed 3 hours and 21 minutes before takeoff, he gave orchestrator prompt that clearly defined setup it was told it was running fully offline on single macbook with no network access that its only resources were local files in users slash dev slash work, llama 70b inference server on localhost 8080, and a battery budget of 3 hours and 21 minutes that it should process task queue line by line, draft outputs, run local evals, save artifacts into done folder, create checkpoints every 12 tasks, and stop only when the queue was empty or battery dropped below 5 percent so system had full awareness of environment it was operating in it knew there would be no outside connection for next 11 hours it knew memory and battery were limited and it knew human would not step in until plane landed system works in single loop it pulls task from queue, sends it through inference, saves output, and writes checkpoint. then it repeats same process again and again when battery drops below 5 percent, orchestrator pauses on its own, waits for laptop to switch to backup power bank, and then resumes from the latest checkpoint these are kinds of entries it was writing into the log during flight saved context checkpoint 8 of 12 with pos min 488 pos max 50118 and size 62.813 mib restored context checkpoint with pos min 488 and pos max 50118 prompt processing progress n tokens 50 out of 60818 task 37016 done tps 71 s tokens text saved to users slash dev slash work slash done slash proposal_westside.md outside window there were clouds and blue sky with no wifi anywhere on tray there was just one macbook, terminal open across two screens, and an inference server running on localhost from everything i have seen, this is one of cleanest offline ai workflows of past year 11 hours in air 0 dollars spent on wifi and full client queue finished before landing

@ayrton_df 44m sushi trade is still undefeated😅🤣

in 2014, 19 year old mit student raised 500000 dollars and gave every undergraduate on campus 100 dollars in bitcoin. alot of them spent it on sushi. that sushi would later be worth 44 million dollars. jeremy rubin was sophomore studying computer science when he ended up in legal fight with new jersey attorney general, who called him hardcore hardened cyber criminal. his offense was creating bitcoin mining program that had actually won an innovation award at hackathon. he was eventually cleared, but bigger thing he took away was that even many students at mit had never heard of bitcoin. so he raised 500000 dollars from alumni and bitcoin supporters, and in october 2014 gave every mit undergraduate 100 dollars worth of bitcoin, around one third of coin when btc was trading near 336 dollars. 3108 students signed up. experiment became first academic study of bitcoin adoption. within two weeks, one out of ten students had already cashed out. by 2017, one out of four had sold. students who simply forgot about it ended up doing the best. most popular place to spend bitcoin was one sushi spot called thelonious monkfish. it was only place on campus accepting bitcoin at time. one student said about half people he knew used their full 100 dollars on sushi. another spent hers on two sushi dinners and forgot rest was still there. in 2021 she checked wallet again. that original 100 dollars in bitcoin had grown to 13000 dollars. if every student had just held onto it, full 500000 dollar experiment would be worth more than 110 million dollars today. as one student put it, most of us thought it was kind of joke. some of smartest students at one of top engineering schools in world were handed free bitcoin in 2014 and used it on sushi. ones who forgot they even had it turned out to be real winners. sometimes best investment strategy is just losing access to your wallet for few years

@kenso_yapys gta5 forever🫀🫂

17 year old guy from china ordered raspberry pi zero for 4 dollars, diodes for 2, and small screen for 6 on aliexpress. 12 dollars total and parts showed up at his door he put it together himself, connected it to fivem server, and with zero coding experience wrote his first lua script through claude in just 90 minutes he simply explained what he wanted, and claude generated full thing for him including job system, economy, ui panel, and config file so server owners could customize it without editing code he listed it on cfx marketplace for 89 dollars, then raised it to 189, and later bundled everything together for 389 developers on tebex usually take up to 90 days to reach 5000 euros month, but he got there even faster because claude built in 90 minutes what would normally cost 50 dollars an hour from developer gta 6 is coming in 7 months, and rockstar is building creator economy directly into game. gta 5 made 8.6 billion over 13 years and paid creators nothing because there was no native platform. gta 6 looks like completely different setup 12 dollar hardware 90 minutes first script on marketplace

This was all streamed from Bitcoin using Ordinals. The 3D world supports real-time multiplayer interactions, and decentralized land ownership. All this is on-chain on Bitcoin, and as immutable and secure as Bitcoin itself. The weather in the world reacts to Bitcoin fees, clouds are linked to transaction volume, and time in the world is determined by block height. Makes sense for a Bitcoin world. Real-time multiplayer interactions is one of the newest features. This is particularly difficult to do because Bitcoin block times are ~10 minutes. Bitcoin has been around for 17 years, and we're now seeing fascinating new things being built on top of it.

@ben_w3bber 🔞🔞🔞

imagine being lazarus group you hack kelp dao for 292m in rseth instead of dumping it, you post it as collateral on aave and borrow 190m in clean eth against it aave loses 8b in tvl within 48 hours in what becomes the first real bank run in defi the arbitrum security council freezes 71m, which ends up being the only recovered money the remaining 175m gets routed through thorchain, with 494k paid in protocol fees for the privilege then it gets turned into btc and scattered into utxo dust across thousands of wallets after that it moves to tron and gets swapped into usdt chinese otc desks gather the flows and settle them through unionpay outside swift and outside sanctions the cash reaches pyongyang and helps fund the missile program 7 days, 9 protocols, everything happened fully in public on chain, and nobody stopped it. defi in a nutshell

most major global indices are starting to show weaker momentum on higher timeframes while sitting at or near all time highs. i am still in cash apart from my long term etf holdings (almost) since moving to sidelines in october, i still have not seen anything that makes me want to start buying back in. right now i do not see strong reason to enter this market

@stoner_races farcaster fork wars plus airdrop is good

farcaster is going through its own bitcoin cash style split as former farcaster developer cassie is launching separate fork called hypersnap and good part is that there will be token and an airdrop you can check your hypersnap token allocation based on your farcaster activity here farcaster.xyz

@ayrton_df congratulations on your purchase!

better late than never🥳

@kenso_yapys real yield has to beat bonds😎

these are kind of protocols we want to find defi yields have tightened lot while risks are still easy to see, that has to improve if we want meaningful growth we are looking for protocols that run their own strategies along with vaults or yield assets that can generate returns above treasury bonds we are also open to strategies that go through drawdowns on the way since that works well with our product if you are building loops pt strategies or money market protocols and trying to grow credit, we want to hear from you if your protocol matches this and you want to discuss integrations, send message or drop comment under post

@ben_w3bber hmm🤔

9 more protocols were hit after 280m drift exploit cow swap got hit through frontend and dns takeover hyperbridge was exploited through forged messages 1b tokens were minted and price went to zero bybit faced 1b exploit attempt it was stopped in time kucoin saw 9.5m laundered kraken dealt with an insider extortion attempt more than 2000 accounts were exposed dango had a smart contract bug 410k was drained silo v2 suffered an oracle exploit 392k was lost bsc tmm got hit by reserve manipulation 1.67m was lost aethir and subquer had access control issues combined losses reached 480k in total around 45 protocols have been hacked in 2026 overall losses are already above 450m so what is weakest point right now code people or system itself

@foodolla dream...

imagine leaving the wagecuck lifestyle to run a b&b… couldn’t be me

a drunk Chinese friend who works as a quant slipped me a napkin with 8 words on it we were at a rooftop in tel aviv. he was three cocktails past the professional filter. i mentioned prediction markets. he laughed. grabbed a napkin from the bar. scribbled in english. "disposition. category. velocity. cross-market. news. volume. whale. theta." i asked what it means. "eight signals. every mistake on these markets is one of them. we run detectors in parallel. enter only when three or more fire" i asked how long it took to build. "two years. fourteen quants" i asked if i could just do it myself. he laughed harder. "with Claude? tonight probably" i don't think he meant to say that part. he looked at the napkin like he regretted writing it. too late. i pocketed it. went home. opened Claude at 11PM. typed the eight words. "build a parallel detector grid on prediction markets. 8 signals. enter only when 3 or more fire simultaneously" Claude needed a backbone. i gave it: github.com/warproxxx/poly… 610 stars. market making bot. order book execution on both sides. position sizing. the foundation for anything that needs to place real orders fast. terminal was live by 2AM. the radar lights up 14 times per hour. most signals fire alone. rare ones stack. last week: > BRICS summit - 74c. 4 signals. +24c in 3h. > Artemis launch - 63c. 5 signals. +33c in 5h10m. > Helium-3 demo - 68c. 3 signals. +14c in 48m. > Waymo public - 67c. 4 signals. +8c in 6m. sentiment feed reads 4 platforms: - X/CT: 56% bearish - Reddit: 53% - Telegram: 17% - Discord: 75% when sentiment diverges from flow pressure by 20%+ the contrarian signal fires. 83% win rate on those. cross-market delta catches gaps: BTC 92K vs Kalshi 7.7c Fed cut vs PredictIt 7.2c AAPL>180 vs Manifold 5.7c oracle delta tracks NOAA, AP, CoinGecko, Reuters. when a feed moves before the market adjusts, the bot enters on the lag. 6 wallets copied by dominant category: > 0xe41f8...b2 - $483 profit. 83% wr. > 0xa22c3...d9 - $890. 67%. > voltwhale - $593. 55%. > 0xf88d2...c4 - $830. 73%. 412 trades. 76% win rate. avg hold 5h22m. best $232. worst -$37. 9 consecutive wins. +$12,640. from a $900 seed. 4 weeks live. copytrade setup: @1743116" target="_blank" rel="nofollow noopener">kreo.app/@1743116 i haven't seen him since that night. the napkin is still in my drawer. i don't think he remembers writing it.

only during boring market can i let myself rest🤠 rn im locked in on @MEXC and ai, left @Polymarket unattended for week lol have good weekend @everyone!