Benny da Bull (@DynamicalBull) header
Benny da Bull (@DynamicalBull) profile picture

swinging shit - turning 25k to 1M roth IRA

Joined September 2021
Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)Photo by Benny da Bull (@DynamicalBull)
Benny da Bull retweeted
Supervised fine-tuning can teach an LLM to write code without running that code! Given a prompt and a demonstrated solution, the model scores each answer token using the prompt and earlier target tokens. The loss drives a weight update that makes those demonstrated tokens more likely. The training signal comes from the example, so good demonstrations matter.
1
3
14
620
Benny da Bull retweeted
The British prime minister is calling for Russell Brand to be punished for mentioning Jesus. Now you know what his entire “rape” trial was about. Megyn Kelly explains. 0:00 Russell Brand's Trial 7:45 How Long is Brand's Sentence? 11:18 The Netflix Documentary 29:24 The Corruption of the British Government 37:47 Has the US State Department Weighed in on This? 40:34 The Anonymous Accusers 44:52 Where Are the Real Journalists?
🤝 Paid partnership
508
1,715
83
7,094
415,328
Benny da Bull retweeted
One neuron, by the numbers. Inputs 0.2, 0.5, 0.43 times weights 0.4, 0.26, 0.3 give 0.08 + 0.13 + 0.129. Add bias 0.1: z = 0.439. Activation: σ(z) = 1/(1 + e⁻ᶻ), so σ(0.439) ≈ 0.608. That number is passed to every neuron in the next layer. Training picks the weights w that minimise the average loss (1/n) Σ ℓ(h(xᵢ; w), yᵢ), by stepping down the loss surface along −∇L. Learning is weighted sums, a squashing function, and gradient descent.
5
81
2
443
11,762
Benny da Bull retweeted
When the $VIX is high it’s time to buy formula
Usually once or twice a year, when the $VIX spikes to extremes, I tell my followers on instagram-regular people that usually sit on cash with no market knowledge to start buying. I did it during COVID. I did it again when the market dropped ~15% on tariffs. This year, by luck, I nailed the exact bottom. I don’t always catch it perfectly, but I’m usually close enough. To simplify things (so they don’t need me), I built a simple framework: If the $VIX is above 30 and the market is down more than 8%, invest 50%-then scale in until fully invested. I’ve attached a simple table below that shows if you have $10,000 in cash exactly how much to add at each step-so anyone can follow it without overthinking.
23
17
1
323
44,820
Benny da Bull retweeted
a while ago i came across learn-inference.com/ and absolutely loved the idea of making technical books more interactive. got inspired to build something similar for one of my favorite technical books, Designing Data-Intensive Applications (DDIA). a first look at what i have been building:
27
40
615
18,962
Benny da Bull retweeted
今年の共テ物理これらしい
192
1,757
272
16,314
836,724
Brutal selloff in $QQQ, but bulls may have a shot at bottom today. QQQ gapped up over 1.5% but looks set to close down. Since 2010, QQQ was higher 5 days out in all 7 cases, with no further downside in 6. There were few such days around Dotcom and GFC, but they weren't so kind.
14
32
3
239
36,914
Benny da Bull retweeted
Multi-agent systems often behave like a mob, not like a coordinated team. We are releasing Multi Agent Coordination Bench: a benchmark requiring agent coordination, collaboration and communication. Right now, agents benefit from parallel exploration, but their coordination and communication is very poor. A team of eight agents does better than a single agent but often due to random exploration, while their coordination has significant room for improvement. Our tasks include agent teams trying to execute a mission to Mars, fight zombies, grow a company from seed, and more. We categorize our tasks by two properties of the environment: if the agents share the same information and actions (Shared vs Distributed), and if they pursue the same goal or not (Collaborative/Strategic). This taxonomy will be useful for future research, to think about different multi-agentic problems across distributed and strategic settings.
We’re releasing Multi Agent Coordination Bench, a benchmark of twelve natively multi-agent long-horizon tasks that measure the coordination capabilities of frontier models. We find that current frontier agents coordinate poorly. A team of eight agents beats a single agent, but the gain comes from having more agents, not from coordination. Of Opus 5.5, GPT-6.1 Sol and Gemini 4 Argon, only Opus shows signs of a gain from coordinating.
37
57
11
498
38,562
Benny da Bull retweeted
1/ AI agents can now accomplish many research tasks. But tackling frontier problems requires deciding which questions matter and when to trust results. We're excited to introduce DiscoveryAgent to empower scientists in their pursuit of rigorous, open-ended discovery 🧵
31
61
8
424
36,724
Benny da Bull retweeted
Elon Musk says $NVDA made a mistake removing Jensen Huang confirming they’ve spoken privately and that $SPCX would hire him himself “who wouldn’t?” He compared Jensen to “Michael Jordan at his peak” saying Nvidia essentially kicked its most valuable player off a winning team.
94
277
24
3,169
223,906
Benny da Bull retweeted
I copy any animation I see now, these 18 GitHub repos are the reason a f*cking gold for Fable 5.2 Copy. Style. Paint. VFX. Beat. Captions. reanimate → rebuild any animation you see, frame by frame github.com/NiazMorshed2007/r… motion-design-skill → timing and easing like a trained animator github.com/LottieFiles/motio… mg-styles-15 → 15 motion styles, one prompt each github.com/Vincentwei1021/mg… huashu-art-motion → paintings that move in 35 art styles github.com/alchaincyf/huashu… anidoodle → doodles and loops, all in code github.com/alexgreensh/anido… cinetic → launch films straight from code github.com/Leonxlnx/cinetic motionmaxxing → turns AI slop into designer-level motion github.com/Tejashmakwana/mot… Manim → turns formulas into smooth animation github.com/ManimCommunity/ma… blender-skills → pro Blender tricks your agent can use github.com/EL4CTEO/blender-s… three.quarks → particle effects that make every scene pop github.com/Alchemist0823/thr… react-postprocessing → the cinematic finish on every frame github.com/pmndrs/react-post… Tweakpane → sliders for every tiny detail github.com/cocopon/tweakpane EffectCraft → fast compositing, built in Rust github.com/storytold/effectc… Openverse → free sounds and music for every scene api.openverse.org/v1/ librosa → finds the beat so your motion follows it github.com/librosa/librosa WhisperX → word-perfect captions, timed to the millisecond github.com/m-bain/whisperX Remotion Skills → official video skills for any coding agent github.com/remotion-dev/skil… FFmpeg → turns your animation into a real video file github.com/FFmpeg/FFmpeg Copy → style → move → sound → ship. Bookmark this, then open the guide below and animate something nobody can scroll past ↓
Fable 5.2 can make a film from one prompt Most of those films will be forgotten in a day A longer prompt won't save them, the problem is how you direct it I broke down 9 lessons from 40 finished films, with every prompt, setting and number ↓
40
324
8
3,011
185,761
working on our empty states @mastra each one has custom error animations too, although we hope our users never have to see them.
10
8
328
7,742
Benny da Bull retweeted
This is such a brilliant framing for the Huang/Nvidia story: “No matter what you are selling, you always want to commoditize complements to your product.” If you don’t understand what @Devon_Eriksen_ means by that, it’s a must read.
No matter what you are selling, you always want to commoditize complements to your product. That might sound like gibberish until I explain the terminology. To commoditize something is to make it into commodity. A commodity is a product that's available from lots of different brands, they are mostly the same, and they mostly compete on price. A compliment to your product is something that people use with your product. Especially if they need it to use your product. In other words, if you sell cars, you want gasoline to be cheap, plentiful, and all pretty much the same. And if you sell gasoline, you want lots of brands of cheap cars out there that are pretty much the same. If you sell electric cars, you want everyone to have rooftop solar panels. If you sell rooftop solar panels, you want everyone to have electric cars. But what if you sell floating point processors that run AI models? Well, you want AI models to be a commodity, don't you? And in the software world, "commodity software" usually mean open source. And what if you sell AI models as a subscription service? Well, you sure as hell don't want all of these open source AI models floating around commoditizing your product, do you? You want the opposite... a nice fat regulatory moat around AI software, while dozens or hundreds of chip-fab companies make GPUs off open-source designs. So Anthropic and NIVIDA are, by necessity, business partners. But they have opposing goals. One is naturally against open source AI, and one is naturally for it. So, now, let's think about what the accusations against Jensen Huang are. 1. He cheated on his wife. Red herring. Immaterial. Lots of powerful men cheat on their wives, especially on the mid wives they picked up before becoming powerful men. No one cares except retarded social media midwits who just react to whatever the matrix shoves under their noses, like a canary pecking at a mirror, without ever asking why the matrix shoved this particular tidbit at them. 2. He was doinking a Chinese intelligence asset. I don't know if this is true or not. Sources differ. But it does occur to me that plenty of our politicians are doinking Chinese intelligence assets and no one cares. So, again, why did the matrix shove this particular tidbit at you? 3. He is too cozy with China. Here, at last, is the rub. What aspect of China? And why? Was he passing NVIDIA's tech secrets to mainland China, because Taiwanese people just love the CCP so much? Or was he trying to provide certain types of support for the development of open source AI models? So, how should you feel about this? Well, I won't decide for you, because it depends on what you want. But I will tell you what the real issue is. Is more important (and possible) to keep cutting-edge AI out of the hands of the Chinese, so they can't compete with America? Or is it more important to have open source AI models freely available, so you and I can compete with Sam Altman and Dario Amodei?
29
63
4
1,040
106,128
Benny da Bull retweeted
The more you know about this case, the more you realize the Epstein class will do literally anything to keep power.
Everyone has an opinion about the Russell Brand case, but few have the facts. This is the other side of the story.
774
4,669
115
26,257
1,231,925
Benny da Bull retweeted
introducing /prune-tests find and remove tests that: - change whenever the code does - check deleted features are gone - can never fail video by opus 5.5
>hey opus please go remove <thing> opus: <thing> is now removed. I added 4000 lines of tests asserting that <thing> no longer exists so that it can't come back without us knowing.
82
115
16
4,001
209,536
Benny da Bull retweeted
Introducing Lemonade OS Midnight. Our autonomous software factory, built to take software from idea to production with minimal human intervention. It doesn't just write code. It writes our code. The vast majority of it. And yes, it works while we sleep.
60
74
25
972
103,690
Benny da Bull retweeted
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation. Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
61,144
121,720
24,504
1,174,484
58,825,907
Benny da Bull retweeted
Stan Weinstein on TV during the 1987 crash: • It didn’t come out of nowhere. The technicals turned negative weeks earlier • The market broke its 30-week MA • Don't buy the first bounce, sell into it if you are still invested Nearly 40 years later it’s still the same playbook
15
74
8
713
49,703
Benny da Bull retweeted
Introducing Orion, a foundation model that predicts where you look. Orion finds twice as many UX bugs as frontier models, cuts computer-use latency by 31%, and even won a drone race.
68
113
18
2,352
85,846