Загрузка видео...
Не удалось загрузить видео
อึ้ง น้องเรย์บอกเองเลยว่ามารายการนี้ตื่นเต้นมากไม่เหมือนเวลาไปออกวาไรตี้อื่นๆเพราะเห็นซานะซอนเบนิมตั้งแต่ในรายการ sixteen มาก่อน แล้ว sixteen มีตั้งแต่ 2015 ก่อนทไวซ์เดบิวต์น้องเรย์อายุ 11 อะ ยังไม่มาเป็นเด็กฝึกด้วยซ้ำ เห็นพี่เค้ามาตั้งแต่เด็กจนวันนี้ได้มานั่งในรายการด้วยกันแล้ว
Комментарии: 0
Нет доступных комментариев
Здесь появятся комментарии из оригинального поста
![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!](https://image.24vids.com/tw-2081382845243691256/ext_tw_video_thumb/2081382802667249664/pu/img/n2gZcnKRYe9G781o.jpg)
![🚨It’s 3:20 a.m. in Tehran. Here are my 16 key observations on the current state of protests in Iran (January 13). 1⃣ Sixteen days of nationwide protests—from December 28 to January 12—mark the largest mass demonstrations in Iran’s modern history. As of this hour, there are no verified reports of street protests today. 2⃣ According to multiple reports, at least 12,000 people have been killed—a figure that can only be described as a massacre. Observers warn the real number may be significantly higher. 3⃣ Iranian diaspora media—including Iran International, Manoto, and BBC Persian—shifted coverage today. After days of cautious reporting to avoid spreading fear, they began documenting the scale of the regime’s crimes. Iranian society is in shock; emotions are raw and volatile. 4⃣Regime media struck a triumphalist tone, citing yesterday’s pro-government rally and claiming that “with the wisdom of the people and security forces,” Iran’s enemies were defeated—just as in the recent 12-day war. Tasnim, an IRGC-affiliated outlet, ran a full-page headline: “Iran won its second war with Israel in one year.” 5⃣Opposition voices counter this narrative, pointing to the 1979 Islamic revolution’s 398-day trajectory and arguing that revolutions are marathons, not two-week events. They describe the current path as irreversible and insist this is not the end of the game. 6⃣ Regime media reported that, over the past two days, security forces raided and destroyed dozens of alleged “armed riot cells.” Police officials claimed hundreds of arrests in Tehran, while provincial authorities cited widespread detentions nationwide, vowing to pursue protesters “to the last person.” 7⃣ Tehran’s prosecutor announced that cases accused of “moharebeh” (punishable by death) are being fast-tracked, with indictments already issued or imminent. 8⃣The regime’s foreign minister told Al Jazeera that Iran would not abandon “justice for people killed by foreign agents,” implicitly framing January 8—the first day of the Crown Prince’s call to protest—as “Day 13 of a U.S.–Israel war against Iran.” 9⃣ Senior officials escalated rhetoric: the president’s executive deputy branded the Crown Prince Reza Pahlavi “unpatriotic” for appealing to Trump; Khomeini’s grandson—without evidence—claimed protesters beheaded four people “ISIS-style.” Reformists and hardliners alike amplified coordinated, baseless accusations that protesters were beheading people and burning citizens alive. 🔟A regime figurehead claimed that two-thirds of the dead were “martyrs” killed by foreign actors—a familiar pattern over 47 years of blaming the U.S., Israel, and “enemies” for domestic unrest. At the same time, the regime refuses to hand over the bodies to their families in many cases. 11. Limited phone connectivity allowed Iranians abroad to receive harrowing messages from inside the country: accounts of relatives killed, questions about why the U.S. and Israel did not intervene, and pleas for international support amid mass arrests and fears of darker days ahead. 12. Trump, across social media and interviews, promised accountability for the regime’s crimes, warning that executions would trigger decisive action and emphasizing that the ultimate objective is “to win.” This messaging has sparked renewed hope. 13. At Trump’s request, Elon Musk made Starlink subscriptions free in Iran. An estimated 50,000 Starlink terminalsexist in the country, many previously offline due to cost. Meanwhile, state TV declared social media an “enemy battlefield” and said the internet will remain shut until “security conditions normalize.” 14. As limited connections return, families abroad are learning—often belatedly—of loved ones killed. One X user wrote: “I wish the phones hadn’t connected. They killed my cousin’s only 24-year-old son at his doorstep, in front of his father… I’m losing my mind.” [2nd pic] 15. New videos from January 8–9 are emerging with delay due to blackouts, revealing protests far larger than previously estimated. Many say January 8—the first day of the Crown Prince’s call—will be discussed for years. A journalist relayed from Tehran: “Everyone came out with their families. The regime saw the crowd—and chose mass killing.” 16. Outlook: The situation is extremely fragile. Street protests may be paused today, but public anger and shock are unresolved. Voices inside and outside Iran converge on one message: increase international pressure and demand global intervention. 2026 is a decisive year. The drivers of revolt remain—and in worse form. Resistance is likely to continue, possibly beyond traditional street protests. P.S. The first video dates back to January 8 in Mashhad and has just been released.](https://image.24vids.com/tw-2011225602318102957/amplify_video_thumb/2011224594229637122/img/QJMhAt8CtTjEKvr8.jpg)