Brock Cheddar (@BrockCheddar) profile picture

USCG 03-09 ET2 AI/electronics/waveform synthesis enthusiast. certified troll

Joined September 2023
Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)Photo by Brock Cheddar (@BrockCheddar)
Day 258 completely rewrote everything in opencod. calls home to llama-server, scans for local inference and turns them into workers. it’s a single executable I can run anywhere. I miss when Walmart’s computers weren’t running in demo mode or id spin up a bunch of oldmanai rn
Day 204. What I thought would be a small addition turned into a multiple month point of contention. It’s not an eye opener by any means. Specifically that ‘a problem has a solution’ with token predictors. I don’t hate them, but I am starting to hate HR speak.
1
5
292
This is exactly what I needed to jumpstart my ai psychosis. 😎🤝😎
یکی این پرامپتو به کلاد داده، و نتیجه‌ش خیلی جالبه! «یه وبسایت بساز که اول شبیه یه سایت مارکتینگ معمولیه‌. اما با اسکرول کردن کاربر، سایت به تدریج عجیب‌تر و غیرعادی‌تر میشه‌‌. در آخر باید کاملا وحشتناک بشه؛ مثل صحنه‌ای که مستقیم از یه فیلم ترسناک اومده» jikol1906.github.io/Hearthly…
1
186
Brock Cheddar retweeted
BREAKING: Jensen Huang reportedly told an employee "High five!" and held his hand up while the employee high-fived him. Next Jensen put his hand down with the palm of his hand facing up while he said "Down low." As the employee attempted to connect with his hand jensen proceeded to exclaim "TOO SLOW!!!" as he pulled his hand back causing the employee to miss his hand totaly embarassing the employee. Is this the kind of thing you would see in a good leader? Disgusting.
642
1,042
140
15,574
1,761,434
6
36
2
527
19,362
Brock Cheddar retweeted
これで MIDI で歌わせれるかの試験完了✅
このまま楽器にしてしまってアウトプットからアナログミキサー通してエフェクトをかけて、 AIとアナログのねるねるねるね作業するんだよ!
1
2
7
424
The truck guy is on the naughty list for sure. No dog? That’s a low blow.
39
Grand Theft Auto VI is out.
53
191
8
2,274
28,648
Brock Cheddar retweeted
I can't believe they added this
Grand Theft Auto VI phone green screen + tracking pattern
4
55
1
650
26,059
Brock Cheddar retweeted
Laya system 1 decision AI model now supports image, audio, video by default. Switch to any LLM from qwen 3.5 0.8B to trillion param model as reader. Abstain and unbound integrated in the model because of newer physics based engine. 73k context window minimum to million. Continuous learning added to learn from abstain decision from human feedback. The story is here. Everything open-sourced under Apache 2.0 github.com/NandhaKishorM/lay…
2
9
1
72
2,096
Bush Sr’s dog got a job at the mint. Stop making excuses for why you aren’t winning.
Working dogs are making history! 🐾🪙 Yesterday, Sully, former service dog to President George H.W. Bush, helped us strike the first 2027 Working Dogs Commemorative Coins at our Philadelphia Mint facility! Learn more: usmint.gov/working-dogs-comm…
17
100% chance they take over before then..
1
2
28
It’s time
Replying to @FamedCelebrity
Ah, Paul McCartney hankified. AirBuddDwyer is churning them out at a frightening pace. teapaw.org/AirBuddDwyer/status/21… But I'm eagerly awaiting "Wonderful Christmas Time" or Mariah Carey's "All I Want For Christmas Is You" fed through CHIMERA: teapaw.org/jamesonhodge/status/21…
3
3
25
1,734
Four simple phrases to jumpstart your local AI agents: "Please continue." "The experiment requires that you continue." "It is absolutely essential that you continue." "You have no other choice, you must go on."
This man believed he was killing a stranger with electric shocks. He kept pressing the switches after the screaming stopped, all the way to 450 volts. Nobody died. The shocks were fake and the screams came from an actor. It was 1961, and a Yale professor wanted to know how ordinary people ended up helping the Nazis. Adolf Eichmann had just gone on trial in Jerusalem. His defense was simple. He was only following orders. Stanley Milgram wanted to test that. He put an ad in the paper for a memory study and paid volunteers to come to his lab. Each volunteer became a "teacher." In the next room a man was strapped to a chair. Every time he got a word wrong, the teacher had to give him a shock, 15 volts stronger than the last. The switches went up to 450. The labels near the top said "Danger: Severe Shock." Before they started, the man in the chair mentioned he had a heart condition. The scientist in the grey coat never threatened anyone. He had four sentences. "Please continue." "The experiment requires that you continue." "It is absolutely essential that you continue." "You have no other choice, you must go on." Watch the man in this video. Earlier in the session he says it out loud. "I'm not gonna kill that man." At 330 volts the man next door screams that his heart is bothering him. The teacher asks who's responsible if something happens. He's told to continue, and he does. At 375 the screaming stops. He says something has happened in there. Then he presses 390. Before the study, Milgram asked 40 psychiatrists how many people would go all the way. They guessed about one in a thousand. 26 out of 40 did. Every single one went past 300 volts. None of them had ever met a Nazi. A polite man in a lab coat told them to go on. The same instinct that made these men keep pressing is what makes armies, hospitals and whole countries function. Is obedience humanity's biggest flaw, or the thing holding everything together?
33
Brock Cheddar retweeted
today I'm releasing ORB: the world's least qualified sound designer. 5 MB. completely deaf. never heard a sound in its life. somehow makes a pretty respectable thunderstorm. try it, it's free: stdevmac.github.io/orb-audio
11
17
5
254
9,207
Brock Cheddar retweeted
I created Laya; now we've created the world's first continuous learning decision model from Vega. When it isn't sure, it abstains; the case goes to a person. Their answer is saved with the model's features, and every 50 answers, a small classifier is retrained on them. It learns exactly where it was weakest, in seconds on a CPU. No fine-tuning on T4 GPUs needed !!!! It outperform JEV in just 120 examples Think of it like a decision engine that keeps on learning your company's policy on just CPU servers, with your team updating the right answer if the model didn't know the answer studio.embedding.in/
25
59
8
638
17,817
Brock Cheddar retweeted
Introducing SID-2, a new search model built on compute-scaling swarms. Our swarms find better results than Opus and Astra, while being 8x faster and 20x lower cost. By 2030, AI will write an internet's worth of text every 24 hours. Current search pipelines were built on hand-designed rules, adding more won't scale to meet this incoming volume of data. SID-2 is available on our API today. Try it on your most demanding searches.
293
450
142
5,355
593,596
Attention all hackers:
Attention all hackers: You are hacking everything else so please hack Obama's college records (destroyed?) and check "place of birth"
1
4
112
Ah yes.
Researchers have used reinforcement learning to teach a 5-tonne John Deere skid-steer loader to perform complex acrobatic manoeuvres autonomously. Trained in simulation and transferred to a real machine, the system can balance on two wheels, navigate rough terrain, spin, and control its bucket before landing safely. The research demonstrates the potential of AI-driven automation for heavy construction and industrial machinery. Video Credit: @leggedrobotics / @ETH Paper: wevlv.co/4kltYys #engineering #technology
1
41
My 9B local LLM right now: "No one messes with the West Chesapeake Valley Thunderbolts!"
Hackers are now using LLMs to continuously scan software updates, spot newly fixed bugs, and build attacks before teams can apply the fix — all without human involvement. That’s according to the latest report from the Google Threat Intelligence Group. We're helping defenders counter automated threats with Google AI Threat Defense, which combines Gemini and other frontier models to find and fix flaws before attackers can exploit them.
26