Video wird geladen...
Video konnte nicht geladen werden
๐๐๐ง๐จ๐ฉ ๐ฉ๐๐ฉ๐ก๐ ๐๐ฃ ๐ฅ๐ง๐ค๐๐ง๐๐ข ๐๐๐จ๐ฉ๐ค๐ง๐ฎ ๐๐ค๐ง ๐ฉ๐๐ ๐๐๐ก๐ซ๐๐ง ๐๐๐ฌ๐ ๐จ๐ Lincoln SW goes on an 11-0 run sweeping Papillion-La Vista to take Class A๐
37,564 Aufrufe โข vor 2 Jahren โขvia X (Twitter)
1 Kommentare
Jim Polejewskivor 2 Jahren
@bigredmm32 My favorite part of watching it was Jessica standing there watching her team celebrateโฆTom Kelly like moments
![Used Product Commercial Skill to create this video on FlovaAI FlovaAI_Japan prom SHOT 1 [0:00โ0:03] โ 3s, real-time Extreme close-up: a single perfect golden-brown fried mandu (Korean dumpling), triangular fold, sitting on the railing of a 10th-floor officetel balcony in Seoul. Bright midday sun catching every crisp fried edge, steam rising gently. A small dish of soy-vinegar dipping sauce sits right beside it, close to the mandu. SHOT 2 [0:03โ0:05] โ 2s, real-time A pigeon lands on the railing a foot away. It looks at the mandu, tilts its head, takes one casual step closer. Its wing brushes the sauce dish. SHOT 3 [0:05โ0:06] โ 1s, real-time The mandu and the sauce dish both wobble from the nudge. They tip together and fall off the edge of the railing. SHOT 4 [0:06โ0:10] โ 4s, slow motion Wide shot from directly across the street at 10th-floor level: the mandu and sauce dish tumbling outward, rotating end over end, catching the sunlight on every turn. Behind them, the building facade โ a tall Seoul apartment block, muted grey-blue exterior, laundry hanging on a few balconies. The mandu is a tiny golden triangle falling against the building. SHOT 5 [0:10โ0:13] โ 3s, real-time Ground level wide shot: a busy Seoul street in midday sun. Pedestrians with shopping bags, a delivery scooter idling at the curb, a tteokbokki/odeng street cart with an ajumma stirring a pot, a dog sleeping in a patch of sun, an elderly man reading a newspaper on a bench. Ordinary day, nobody aware anything is falling. SHOT 6 [0:13โ0:15] โ 2s, real-time Close-up on the sleeping dog: one ear twitches, then both go up. Nose starts working. One eye opens, pupils tracking something falling. The dog slowly rises, never looking away from above. SHOT 7 [0:15โ0:17] โ 2s, real-time Close-up on the cart ajumma: her stirring spoon slows, then stops. She looks up, eyes widening, and slowly sets the spoon down on the cart edge. SHOT 8 [0:17โ0:19] โ 2s, real-time Close-up on the newspaper man: a small triangular shadow crosses his page. He lowers the newspaper and looks up, his glasses catching the sunlight. SHOT 9 [0:19โ0:23] โ 4s, slow motion Wide shot of the full street: the mandu and sauce dish visible as tiny falling shapes above the crowd. Below, reactions ripple outward โ the dog now standing, the ajumma craning her neck, the newspaper man on his feet, the delivery driver stepped off his scooter looking up, a vegetable vendor shielding her eyes with one hand. SHOT 10 [0:23โ0:25] โ 2s, real-time A second pigeon flying past at around the 5th-floor level narrowly dodges the falling mandu, does a startled double-take mid-flight, and veers off. SHOT 11 [0:25โ0:29] โ 4s, extreme slow motion Extreme close-up: the mandu now three feet from the ground, rotating slowly, every crisp fried ridge sharp in the light, a faint trail of steam still lifting off it. The sauce dish falls a foot above it, tilting but staying upright. SHOT 12 [0:29โ0:31] โ 2s, real-time Ground-level close-up: the dog's mouth is open, perfectly positioned. The mandu drops directly into it. The dog closes its mouth. The sauce dish lands upright on the pavement beside the dog with a small click. SHOT 13 [0:31โ0:34] โ 3s, real-time The dog sits down, chews twice, swallows. It licks its nose once, then glances down at the sauce dish, then up at the silent, staring crowd. SHOT 14 [0:34โ0:40] โ 6s, real-time Wide shot: total silence on the street for a beat. Then the dog lies back down in its sunny spot and closes its eyes. The crowd slowly breaks and goes back to what they were doing โ the ajumma picks her spoon back up, the newspaper man sits and lifts his paper, the delivery driver gets back on his scooter. SHOT 15 [0:40โ0:45] โ 5s, real-time Overhead, back on the 10th-floor balcony: the first pigeon is now sitting exactly where the mandu used to be, looking down at the street below. Bright midday sun, crisp shadows. Hold on the pigeon as the shot settles, closing the sequence. #Flovaai #Flovacpp](https://image.24vids.com/tw-2078857994008433022/amplify_video_thumb/2078857695021666305/img/NEYC4PFuvPQ3jkZV.jpg)
![Graph Convolutional Network by hand โ๏ธ ~ 12 steps walkthrough below Graph Convolutional Networks (GCNs), introduced by Thomas Kipf and Max Welling in 2017, are the tool for data shaped like a graph: social networks, recommendations, biological networks, drug discovery, molecular chemistry. I drew and calculated a simple GCN entirely by hand. Goal: run a two-layer GCN, then a small classifier, on a five-node graph, filling in every cell yourself. 1. Given A graph of five nodes, A to E, with edges between some of them. 2. Adjacency matrix (neighbors) Put a 1 wherever two nodes share an edge, in both directions. 3. Adjacency matrix (self) Add 1s down the diagonal, one self-loop per node. That is just adding the identity matrix. 4. Messages Multiply each node's embedding by the weights and biases, then ReLU. Negatives become 0. 5. Pooling Multiply the messages by the adjacency matrix. Each node gathers the messages of its neighbours and itself. 6. Visualize Node A pools [3,0,1] + [1,0,0] = [4,0,1]. 7. Second GCN layer Messages again: weights, biases, ReLU. 8. Pooling again Pool over each node and its neighbours, once more. 9. Visualize Node C pools [1,2,4] + [1,3,5] + [0,0,1] = [2,5,10]. 10. Fully connected layer Weights, biases, ReLU. This time there are no neighbours to pool, just the node itself. 11. Linear layer One more: weights and biases. 12. Sigmoid Squash each score to a probability (โฅ 3 โ 1, 0 โ 0.5, โค -3 โ 0). That is the classification for each node. You have just classified every node in the graph by hand. โ๏ธ The outputs: A: 0 (very unlikely) B: 1 (very likely) C: 1 (very likely) D: 1 (very likely) E: 0.5 (neutral) The takeaway: a GCN layer is two parts. The top part pools each node with its neighbours through the adjacency matrix. The bottom part is an MLP that transforms each node on its own. A transformer layer has the same two parts, with an attention matrix where the adjacency matrix was. Both matrices do one job, mixing across positions: attention over tokens, adjacency over nodes. In my class I call the GCN the transformer's little cousin: a bit more stubborn, because its attention is fixed by the graph rather than computed from Q, K, and V. Draw the two side by side and the resemblance is hard to miss. ๐พ Save this post! #AIbyHand #GraphNeuralNetworks #DeepLearning](https://image.24vids.com/tw-2077396566864773414/ext_tw_video_thumb/2077396555410051073/pu/img/JX9U_Rk-wmgd1S-x.jpg)
