正在加载视频...

视频加载失败

T H E W H E E L P r e - o r d e r n o w. I SEE STARS Watch the official visualizer for “Eliminator” now on the Sumerian Records YouTube Channel and pre-save / pre-order THE WHEEL everywhere. Featuring -- B&W Zoetrope Limited Picture...

26,533 次观看 • 1 年前 •via X (Twitter)

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

U-Net by hand ✍️ ~ 17 steps walkthrough below I consider U-Net as a key milestone in deep learning, the first image-to-image model that really worked! It came out of medical imaging, an unusual place, not from NeurIPS or CVPR or ACL. Now it is the backbone of diffusion models, which you see in almost all modern image generation models. I drew the network as a C so the matrix multiplication flows naturally down. Tilt your head to the right and it is a U again. 🤣 Goal: push a 3 x 16 image down to a 2 x 4 bottleneck and back out again, filling in every cell yourself. = 1. Given = An image of three channels, R, G and B, sixteen pixels wide, and every kernel the network will use. = 2. Convolution 1 = Let us slide the first kernel over the image. Each output is one multiply-and-add over a 2 x 3 window, and the result is the green feature map. = 3. Find the maxima = We circle the largest value in each 1 x 2 window. Circling first is worth the extra step: it is the pooling decision, made before anything is written down. = 4. Max pool 1 = Let us copy those maxima down. Sixteen columns become eight, and half the detail is gone for good. = 5. Convolution 2 = We convolve again with the second kernel, deeper into the contracting path. The feature map is blue now. = 6. Find the maxima again = Same move as step 3, on the blue map. = 7. Max pool 2 = Eight columns become four. = 8. The bottleneck = Let us convolve once more. This is the bottom of the U, a 2 x 4 block that is everything the network kept. = 9. Spread it out = We start back up. The transposed convolution writes each bottleneck value into a wider grid, leaving gaps between them. = 10. Transposed convolution 1 = Let us fill those gaps by convolving over the spread-out grid. Four columns become eight. = 11. The first skip = We copy the encoder's matching row straight across. This is the skip connection, and it is the whole reason a U-Net can recover detail that pooling threw away. = 12. Convolution with the skip = Let us convolve the upsampled features together with the copied ones. = 13. Spread it out again = Same as step 9, one level up. = 14. Transposed convolution 2 = Eight columns become sixteen, back to the width we started at. = 15. The second skip = The encoder's first feature map comes across, the one made before any pooling happened. = 16. Convolution and ReLU = We convolve, then cross out every negative and set it to zero. = 17. Output convolution = Let us apply the last kernel. Out comes R', G' and B', an image the same size as the one we started with. The outputs: R' = [3, 0, 7, 0, 7, 0, 17, 0, 3, 0, 9, 0, 2, 0, 6, 0] G' = [1, 20, 1, 10, 1, 12, 1, 19, 2, 5, 1, 11, 1, 3, 1, 7] B' = [4, 20, 8, 10, 8, 12, 18, 19, 5, 5, 10, 11, 3, 3, 7, 7] Congrats! You just calculated a U-Net by hand. 💾 Save this post!

Tom Yeh

17,510 次观看 • 20 天前

My dad looked at my screen and said what is this, a hacker game? It was a Polymarket bot making $400 while he was standing behind me. He sat down. I explained. Polymarket runs on one equation. Softmax. The same math Claude uses to pick the next word. C(q) = b · ln Σ e^(qi/b) 93% of traders do not know this formula exists. They look at 40 cents and think cheap. My bot looks at 40 cents and calculates the theoretical price is 58 cents. That is an 18-cent edge per share. Two more formulas do the rest: f = (p·b − q) / b. Kelly Criterion. P(H|E) = P(E|H)·P(H) / P(E). Bayes. I gave Claude all 4 and said find every contract the market has wrong. $600 to $10,190. 425 trades. Still running. 61.2% win rate. Wrong 39% of the time. Does not matter. The sizing formula makes sure wins pay more than losses cost. 87% of wallets lose money trading against this math. The 4 formulas are on Wikipedia. The code fits in one file. The only edge: my bot reads them at 3 AM when the market is mispriced and nobody is looking. My dad still calls it a hacker game. But he stopped asking when I am getting a real job. I built the entire framework: Softmax arbitrage detection layer Kelly Criterion position sizing Bayesian probability recalibration Claude integration for autonomous execution 3 AM mispricing scanner The system runs 24/7. Finds where the math disagrees with the crowd. Executes before the edge compresses. No prediction. No gut feel. Just formulas. You only need Claude + device + 1 hour per day. Giving this free for 24 hours. To get it: 1. Comment the word money 2. Like and retweet this 3. Follow me Himanshu Kumar so I can DM you Save this post. Deploy the formula system this week. Start with $600. Scale on evidence.

Himanshu Kumar

13,222 次观看 • 1 个月前

If you’re an investing beginner, you MUST watch this video. If you’re an advanced investor, watch it as a reminder. Peter Lynch is the most successful Fund Manager of all time. He uses these 45 minutes to cover 95% of all of investing! My Key Takeaways: 1. Personal Edge - Look for the fields in which you have a knowledge benefit. Working in an industry, being a customer, all of that is an advantage. 2. The Key Organ for Investing: The Stomach - Investing is not about brains. It’s about having the stomach. “The real key to making money in stocks is not to get scared out of them.” - Peter Lynch 3. Categories - Categories and labels are guidelines, not hard rules. Successful investing is about flexibility. 4. P/E Rule of Thumb - Stocks follow Earnings Fairly Priced: P/E equals annual growth rate over the next 3-5 years. Expensive: P/E extensively higher than annual growth rate over the next 3-5 years. Cheap: P/E extensively lower than annual growth rate over the next 3-5 years. 5. Balance Sheet Rules of Thumb - Is the BS healthy? a) Cash should be higher than Short-Term Debt b) If Cash - Short-term Debt - Long-Term Debt is only 1/4 of Net worth, the BS is decent c) Total Debt should equal 20% of capitalization or less 6. Focus on Stories - Stock prices move with the stories told about the companies. Have a long-term story for every company you own and check if it plays out. 7. Profit from Chaos - A market decline of at least 10% occurs every two years. Pick up your high-conviction bets at a discount when this happens. 8. Forget about Macroeconomics - Focus on business growth, not GDP growth. “If you spend 13 minutes a year on economics, you’ve wasted 10 minutes.” - Peter Lynch

Daniel Mahncke

493,303 次观看 • 2 年前

DLSS 4.5 performance cost on the RTX 4070 Super, tested in #TheLastOfUs, #ArcRaiders, and #RedDeadRedemption2 at 1440p. I added Red Dead Redemption 2 since it’s one of the cases where the new presets really shine, especially in terms of motion clarity. It’s not easy to showcase the visual differences because of the capture, export, and the extra compression from Twitter and YouTube but If the difference isn’t obvious, focus on the tree at the end of the run. The Last of Us Part I Performance mode – CNN (Preset A) → Transformer Preset K: ~3% cost – Preset K → Preset L: ~4% hit – Preset K → Preset M: ~3% hit Balanced mode – CNN (Preset A) → Transformer Preset K: ~6% cost – Preset K → Preset L: ~8% hit – Preset K → Preset M: ~3% hit Quality mode – CNN (Preset A) → Transformer Preset K: ~4% cost – Preset K → Preset L: ~11% hit – Preset K → Preset M: ~6% hit Arc Raiders Performance mode – CNN (Preset E) → Transformer Preset K: ~4% cost – Preset K → Preset L: ~8% hit – Preset K → Preset M: ~2% hit Balanced mode – CNN (Preset E) → Transformer Preset K: ~5% cost – Preset K → Preset L: ~11% hit – Preset K → Preset M: ~4% hit Quality mode – CNN (Preset E) → Transformer Preset K: ~4% cost – Preset K → Preset L: ~13% hit – Preset K → Preset M: ~8% hit Red Dead Redemption 2 Quality mode – CNN (DLSS 2) → Transformer Preset K: ~2% cost – Preset K → Preset L: ~12% hit – Preset K → Preset M: ~7% hit Unlike what we saw with the 3060 Ti, the performance hit on the 4070 Super is much smaller, and this is where the new presets start to make a lot more sense. The hit isn’t that large, and you’ll likely get better image quality compared to Preset K. But Is it always worth it? That largely depends on the game and the target resolution. Generally, I recommend trying the new presets and seeing for yourself, but do this only if you have an RTX 40/50 series GPU.

BenchmarKing

53,027 次观看 • 7 个月前

Vinyl 1- you seem pretty sad for a girl so in love - 'spark in the dark' black vinyl 2- you seem pretty sad for a girl so in love - exclusive 'sticky sweet' hot pink vinyl 3- you seem pretty sad for a girl so in love - exclusive 'static lover' pop-up blue vinyl 4- you seem pretty sad for a girl so in love - exclusive 180-gram vinyl 5- you seem pretty sad for a girl so in love - exclusive 180-gram vinyl (signed) 6- you seem pretty sad for a girl so in love - exclusive 'sticky sweet' hot pink vinyl (signed) 7-you seem pretty sad for a girl so in love - exclusive 'static lover' pop-up blue vinyl (signed) 8- you seem pretty sad for a girl so in love - target exclusive 'hope like snow' white vinyl 9- you seem pretty sad for a girl so in love - target exclusive 'pretty sad' pink vinyl 10- you seem pretty sad for a girl so in love 'never looked so blue' (Target Exclusive 11- you seem pretty sad for a girl so in love - spotify exclusive 'melt with you' starburst vinyl 12- you seem pretty sad for a girl so in love - ‘spark in the dark' black vinyl amazon exclusive 13- you seem pretty sad for a girl so in love - “hope like snow white" - Indies Exclusive 14- you seem pretty sad for a girl so in love - (hmv Exclusive) 15-You seem pretty sad for a girl so in love - Cultura exclu 16- you seem pretty sad for a girl so in love - 'carve our names' picture disc vinyl 17- you seem pretty sad for a girl so in love - apple music exclusive 'feminine intuition' lenticular cover pink vinyl 18- drop 💀 - chewing gum 7" vinyl 19- The Cure - 7” vinyl (exclusive song “Never Do” included) 20-‘the cure’ 12inch vinyl (Blood Records) exclusive CDS 1-you seem pretty sad for a girl so in love 2-you seem pretty sad for a girl so in love (signed) 3- you seem pretty sad for a girl so in love CD - Target Exclusive 4- you seem pretty sad for a girl so in love - (hmv Exclusive) 5- you seem pretty sad for a girl so in love - tour exclusive cd 6- drop 💀 - cd 7- drop 💀 exclusive cd zine Cassettes 1-you seem pretty sad for a girl so in love - exclusive baby blue cassette 2-you seem pretty sad for a girl so in love - baby pink cassette 3- you seem pretty sad for a girl so in love - Urban Outfitters exclusive cassette 4- you seem pretty sad for a girl so in love - baby pink cassette + signed cd art card 5- you seem pretty sad for a girl so in love - baby blue cassette + signed cd art card 6- drop 💀 - cassette 7- The Cure - cassette ((exclusive song “Never Do” included)) Digital 1-you seem pretty sad for a girl so in love - digital album 2- you seem pretty sad for a girl so in love - amazon digital exclusive 3- drop 💀 Standard Digital 4- drop 💀 Singalong Audio 5- drop 💀 Singalong Video Bundles 1- you seem pretty sad for a girl so in love - ‘spark in the dark' black vinyl + t-shirt bundle 2- you seem pretty sad for a girl so in love - spark in the dark' black vinyl + crewneck bundle 3- cd & baby pink cassette bundle 4- lace album cover oversized blue t-shirt, 5- baby pink cassette bundle 6- swing embroidered bundle 7- 'sticky sweet' hot pink vinyl, baby pink cassette bundle 8- album embroidered hat, baby pink cassette & baby blue bundle 9 you seem pretty sad for a girl so in love - baby blue cassette and signed poster bundle 10 you seem pretty sad for a girl so in love - baby pink cassette and signed poster 11- you seem pretty sad for a girl so in love - exclusive cd with letter to FR fans REMIXES 1- drop 💀 karaoke version 2- drop 💀 music video (singalong pack) 3- drop 💀 (taken that eurostar to france) - spotify exclusive 4- drop 💀 (stalked you on the internet) - apple music exclusive 5- drop 💀 (BTS) 6- drop 💀(iTunes limited cover) 7- drop 💀(we can go real slow - slowed & reverbed) 8- drop 💀(most alive i’ve ever been - sped up) 9- drop 💀(you know all the words - isolated) 10- drop 💀(acoustic) 11- “the cure” performance - Spotify exclusive 12- “the cure” bts - Apple Music exclusive

.

130,589 次观看 • 2 个月前

Insane video shows the drastic change in water level and devastation caused by the flooding in Western NC. Jay-Paul Thibault in Burnsville, NC filmed how quickly the water rose and the destruction that followed. Unfortunately this is the sad reality for hundreds if not thousands of families. Please keep praying, keep giving, keep sharing, and keep helping. I'll link to his GoFundMe to rebuild in the next post. 0:02 - The Distance From the River to the Bottom of the Bridge is 22 Feet Thu 9/26 - 8:48 AM 0:09 - "The water's already up about 5 feet I'm guessing... so it's pretty intense." Fri 9/27 - 7:34 AM 1:27 - "This is the road [completely under water] and there's no crossing on 19W." Fri 9/27 - 8:14 AM 1:48 - "I was just told this flood is going to exceed the flood of 1916 which dwarfed the flood of 1977." (The water is at the 1000-year level) 3:46 - "I think I'm going to lose the bridge." Fri 9/27 - 8:27 AM 5:56 - "Well there's not much I can do about it but the bridge is gone. I've got no way over there until we repair that." Fri 9/27 - 8:37 AM 6:28 - "So I'm safe and I'm at the Fire Station. We're gonna wait it out here." Fri 9/27 - 8:49 AM 6:56 - "We've got some trucks that got swept away and this is getting pretty crazy... there's the house that was just refinished and that looks like it could get swept away. I hope it doesn't. That's Jeff's house. I'd say we're close to 20 feet over the regular height of the river." Fri 9/27 - 10:10 AM 8:56 - "So we are now on the roof [on top of trucks in the fire station garage] hanging out watching the river go by and we've got about 6 more feet before it's at our feet here now." Fri 9/27 - 10:42 AM 9:13 - "We are punching our way out of the roof at this point in case the flood continues [cutting a hole in the ceiling with an ax to get to the roof of the fire station]. As you can see, everything in here is floating away." Fri 9/27 - 11:51 AM 9:43 - "Holy sh!t is all I can say [while on top of the roof]. The entire dump station is gone; the building, the trash containers." Fri 9/27 - 12:02 PM 12:08 - "Look at the house. Holy sh!t. That's going downstream... That house is just floating away. Been there for 100 years probably." Fri 9/27 - 3:08 PM 14:35 - "So I found my pickup truck [upside down and completely destroyed]." Fri 9/27 - 3:09 PM 14:47 - "So here's the road, or what's left of it." 16:34 - "All of my building is gone; all the construction I did... everything I own... everything that I built is gone. I don't know how to express this. The whole farm here is gone. This is where I parked my car and we had the bridge and it's just gone. So I guess you start over. I left the goats unlocked and they would've gone up the hill. But everything I've done for the last 2 years is pretty much non-existent."

Michael | Net Lease Sales & Trailer Park Tales

34,676 次观看 • 1 年前

I charge $999 to ask a business owner questions for 45 minutes. Then Claude does the analysis in 5 minutes. I call it the AI Tools Assessment. It finds 3 to 7 off-the-shelf tools that reclaim 5 to 10 hours a week, and it's the front door to upsells from $3,500 projects to $2,000/month retainers. Here's the entire model: 1) The discovery call is questions only. "Walk me through yesterday." "What tasks do you dread?" "Where does work pile up?" No pitching. A free AI notetaker captures the transcript. 2) Claude runs the entire analysis. Paste the transcript, run one skill, and it pulls the pain points and prescribes the tools in about 5 minutes. It catches patterns you missed on the call. 3) When Claude whiffs on a tool, and fill the gaps. Thousands of tools, grouped by industry. 4) The report is 9 slides. Executive summary, effort vs impact matrix, tool recommendations, a 4-day quick win plan, and the financial impact. I open sourced the template free at 5) I go for the close on the review call. Three questions: which of these is most urgent, do you want to DIY or get help, and what's your timeline? 50 to 60% of clients ask you to implement it for them. 6) Process redesign sells for $3,500 with zero automation. One e-commerce client had an 18-step ad workflow. We cut it to 9 steps. Fixed the process, didn't touch AI, charged $3,500. 7) Knowledge systems are $3K builds. A business broker got 400 emails per listing. We trained a custom GPT on the marketing package, and buyers called it the best broker experience they've had. 8) You don't need an audience to sell this. One guy walked into 30 local businesses offering a free 15-minute mini assessment. 5 meetings, 2 clients. The free mini assessment is the hook for every channel. 9) Co-working spaces are the cheat code. Dennis in our community hosted his first free AI office hours this week. 9 people showed up, 2 became warm leads. 10) AI Concierge is the best upsell of all. Two 45-minute calls a month at $1,200 to $2,000. I have 5 clients and my blended rate is about $1,100 an hour at 99.9% net margin. Two things that make this work: 1) Sell the diagnosis before the cure. The $999 assessment is a paid discovery call that qualifies the buyer and tees up every upsell on the menu. 2) High perceived value can cost you nothing. Unlimited Voxer access sells the retainer. In 3 months across 5 clients I've gotten 4 messages. Full breakdown below. watch, implement, make money. (also available on the Build With AI podcast wherever you get your pods)

Corey Ganim

48,942 次观看 • 29 天前