>>Greenp+K9
I found this video helpful for understanding transformers in general, but it covers attention too: https://www.youtube.com/watch?v=kWLed8o5M2Y

The short version (as I understand it) is that you use a neural network to weight pairs of inputs by their importance to each other. That lets you get rid of unimportant information while keeping what actually is important.

>>qwerto+l9
https://www.youtube.com/@animatedai/videos for visibility

>>qwerty+(OP)
Beautifully done. It reminds me of these wonderful 3D animated explainer videos https://www.youtube.com/@animagraffs.

>>CyberD+Dg
The name of the site is misleading. A more accurate one might be "Animated CNN architecture diagrams". You need to know a bit about neural networks to make sense of these images. Watch the accompanying video, and compare with the images here: https://en.wikipedia.org/wiki/Convolutional_neural_network

>>qwerty+(OP)
Nicely designed. Here is another visualizer for CNNs, from research at Georgia Tech:

https://poloclub.github.io/cnn-explainer/

Another link to various visualization tools: https://github.com/ashishpatel26/Tools-to-Design-or-Visualiz...

Another one: https://playground.tensorflow.org/

>>qwerty+(OP)
Shameless plug for a recent blog post of mine where I try to explain ML convolutions, https://jlebar.com/2023/9/11/convolutions.html

>>Uehrek+ea
You are spot-on. In one of his videos, the author mentions that he used Blender's Geometry Nodes: https://www.youtube.com/watch?v=w4kNHKcBGzA&t=190s

>>dpflan+NB
Thanks for sharing! I especially like the "Understanding Hyperparameters" widget in the poloclub article. I built something similar that people might find helpful[1].

[1]: https://static.laszlokorte.de/conv2d/

>>qwerty+(OP)
I made my own animations once upon a time using manim, not as shiny but might be helpful too

https://www.jerpint.io/blog/cnn-cheatsheet/

>>anon_c+cX1
I love the cynicism, but if we’re looking at Median speed, us is still among the best - 11th for broadband, 26th for mobile. Punching below its weight class, for sure, but not exactly worse than developing countries. The only exceptions are Thailand, Hungary, and Romania IMO - props to them!!

https://en.wikipedia.org/wiki/List_of_sovereign_states_by_In...

>>qwerty+(OP)
For interactive articles on specific ai algorithms, checkout Amazon’s mlu-explain:

https://mlu-explain.github.io/

>>dpflan+NB
ManimML: https://github.com/helblazer811/ManimML

"But what is a convolution?" https://youtu.be/KuXjwB4LzSA?si=qwnZMQYJhDxraGc8 https://github.com/3b1b/videos/tree/master/_2022/convolution... https://github.com/3b1b/videos/tree/master/_2023/convolution...

"Convolution Is Fancy Multiplication" https://news.ycombinator.com/item?id=25190770#25194658

>>37953886 :

> Manim, Blender, ipyblender, PhysX, o3de, [FEM, CFD, [thermal, fluidic,] engineering]: https://github.com/ManimCommunity/manim/issues/3362

Manim + O3DE would be neat and probably useful for learning

A Manim Rubik's cube algorithm video: https://github.com/polylog-cs/rubiks-cube-video/blob/main/co...

Manim API docs: https://docs.manim.community/en/stable/reference.html