Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Created a WebGL neural network visualization for the quadruped robot I've been working on. The more red, the higher the activation. Bluer is less activation, while green is a midpoint.

brandon_xyzw's profile picture

Brandon

6,961 subscribers

195,704 görüntüleme • 2 yıl önce •via X (Twitter)

Bilim & TeknolojiEğitim
LIVE

Anya Rossi• Live Now

Private livecam show

10 Yorum

Brandon profil fotoğrafı
Brandon2 yıl önce

The network output is the desired joint positions, and I've connected them visually to the corresponding joint. At the top we have the inputs to the network, mostly the previous actions taken, proprioception, and commands.

Brandon profil fotoğrafı
Brandon2 yıl önce

Also should have said: made with custom shaders and #threejs of course

Jim Fan profil fotoğrafı
Jim Fan2 yıl önce

Fantastic visualization. Great job Brandon!

Brandon profil fotoğrafı
Brandon2 yıl önce

Thank you🙏

Altaha Ansari profil fotoğrafı
Altaha Ansari2 yıl önce

This is amazing 😍

Pseudonym 🦅 profil fotoğrafı
Pseudonym 🦅2 yıl önce

I need more of these. Wen?

Brandon profil fotoğrafı
Brandon2 yıl önce

I’ve wanted to work on NN visualizers for some time, including different architectures. So, stay tuned.

TJ Evarts profil fotoğrafı
TJ Evarts2 yıl önce

Robodisco!

juweek profil fotoğrafı
juweek2 yıl önce

i'm tryna do something similar in p5.js! without building the robot part lol

Zach Forrest profil fotoğrafı
Zach Forrest2 yıl önce

@GalaxyKate Finally, interpretable neural nets

Benzer Videolar

Mnist Neural network visualization in Odin and Raylib. All cubes now have an opacity based on their activation/weight. I think the codebase is almost ready to be made public
0:39

Mnist Neural network visualization in Odin and Raylib. All cubes now have an opacity based on their activation/weight. I think the codebase is almost ready to be made public

Bones

48,202 görüntüleme • 1 yıl önce

Inspired by Andrej Karpathy's practice of tinkering with smaller GPTs to gain intuition, I've been training small neural nets to play this simple 2d game I created. On the left is a visualization of the neural net firing in real time. Built with Claude, Three.js, and WebGPU and using JAX from Google DeepMind. I’m piping the neuron activation into Ableton as MIDI to trigger Serum 2.
0:21

Inspired by Andrej Karpathy's practice of tinkering with smaller GPTs to gain intuition, I've been training small neural nets to play this simple 2d game I created. On the left is a visualization of the neural net firing in real time. Built with Claude, Three.js, and WebGPU and using JAX from Google DeepMind. I’m piping the neuron activation into Ableton as MIDI to trigger Serum 2.

Aaron Lemke

22,307 görüntüleme • 1 ay önce

Implemented a visualization for the neural network today. This is a neural network being trained to learn the snake game in a terminal. It's built in Rust using Ratatui More input details in thread
1:00

Implemented a visualization for the neural network today. This is a neural network being trained to learn the snake game in a terminal. It's built in Rust using Ratatui More input details in thread

Bones

165,655 görüntüleme • 1 yıl önce

Now THIS is the craziest King Tower activation I've ever seen.
0:22

Now THIS is the craziest King Tower activation I've ever seen.

Dcmat

107,722 görüntüleme • 3 ay önce

Individual Neuron: Neural Network A neuron in a neural network performs a weighted sum of inputs, adds a bias, and applies an activation function like sigmoid, ReLU, or tanh, introducing non-linearity. This output helps the neuron learn and represent patterns in the data.
0:54

Individual Neuron: Neural Network A neuron in a neural network performs a weighted sum of inputs, adds a bias, and applies an activation function like sigmoid, ReLU, or tanh, introducing non-linearity. This output helps the neuron learn and represent patterns in the data.

ₕₐₘₚₜₒₙ — e/acc

47,467 görüntüleme • 1 yıl önce

A peek inside the mind of AI through neural network visualization.
0:25

A peek inside the mind of AI through neural network visualization.

Massimo

60,161 görüntüleme • 4 ay önce

I've been quite inactive for a while, mainly because I was busy working on a prototype for a coming event. This is a wireless speaker with POV LED display for music visualization, and I'm pretty sure it's the best POV display you've ever seen!
0:58

I've been quite inactive for a while, mainly because I was busy working on a prototype for a coming event. This is a wireless speaker with POV LED display for music visualization, and I'm pretty sure it's the best POV display you've ever seen!

BD4SUP

338,532 görüntüleme • 2 yıl önce

Just found one of the coolest things on the internet → It uses WebGL for the 3D fluid interactions + neural network to generate the harmonization for the different melodies 🎵 A true masterpiece by David Li
0:49

Just found one of the coolest things on the internet → It uses WebGL for the 3D fluid interactions + neural network to generate the harmonization for the different melodies 🎵 A true masterpiece by David Li

Steven Tey

133,366 görüntüleme • 3 yıl önce

Visualization of a neural network
0:59

Visualization of a neural network

Interesting As Fuck

48,219 görüntüleme • 10 ay önce

#Snootgame An upcoming mod that I've been wanting to make for a while based on the comic I've been working on. #Renpy
0:44

#Snootgame An upcoming mod that I've been wanting to make for a while based on the comic I've been working on. #Renpy

Arsenic (III)

24,925 görüntüleme • 5 ay önce

Pi Network is Edging Closer to Protocol v23 After Most Major Nodes have now been upgraded Pi Network (Pi Network) is officially nearing the activation of Protocol 23 as the majority of mainnet node operators complete a comprehensive infrastructure migration. This critical transition involves upgrading the network from Protocol 22 while simultaneously shifting OS environments from Ubuntu 20 to 24 and databases from PostgreSQL 12 to 16. The rollout requires intensive internal data reprocessing to optimize subsystems for higher transaction throughput and storage efficiency across the global validator set.
0:30

Pi Network is Edging Closer to Protocol v23 After Most Major Nodes have now been upgraded Pi Network (Pi Network) is officially nearing the activation of Protocol 23 as the majority of mainnet node operators complete a comprehensive infrastructure migration. This critical transition involves upgrading the network from Protocol 22 while simultaneously shifting OS environments from Ubuntu 20 to 24 and databases from PostgreSQL 12 to 16. The rollout requires intensive internal data reprocessing to optimize subsystems for higher transaction throughput and storage efficiency across the global validator set.

BSCN

23,439 görüntüleme • 21 gün önce

🚀 If you've notice I've been quite for a while, is because I've been working really hard on something huge! In the past 6 months, I've been tirelessly working on my upcoming #backend development #course that uses #Node.js + #Express + #MongoDB + #TypeScript + #TSOA.
0:33

🚀 If you've notice I've been quite for a while, is because I've been working really hard on something huge! In the past 6 months, I've been tirelessly working on my upcoming #backend development #course that uses #Node.js + #Express + #MongoDB + #TypeScript + #TSOA.

Vandad Nahavandipoor

13,804 görüntüleme • 3 yıl önce

The camera following the swing is some of that neuron activation kind of shit for me
0:30

The camera following the swing is some of that neuron activation kind of shit for me

Etheox

17,344 görüntüleme • 1 yıl önce

I wrote a blog post about my own Python tensor compiler library called TensorFrost that I've been working on for the last 14 months! As a sneak peak here is the last pet project that I've made using it - an implementation of Neural Cellular Automata:
1:23

I wrote a blog post about my own Python tensor compiler library called TensorFrost that I've been working on for the last 14 months! As a sneak peak here is the last pet project that I've made using it - an implementation of Neural Cellular Automata:

Mykhailo Moroz

11,989 görüntüleme • 1 yıl önce

Motorsport is cruel. I've been working on this for over a year, and just when things start working, the engine explodes...
2:20

Motorsport is cruel. I've been working on this for over a year, and just when things start working, the engine explodes...

Pablo Araujo

16,074 görüntüleme • 8 ay önce

Super simple Thorn buff idea: Reduce the activation timer for the Razorbloom. (Not to be confused with the detonation timer after it's been triggered.) Lowering the activation timer allows her to be a reactive area denial op rather than solely a trap op. This makes her more useful and more interactive to play. It will still be just as easy to evade/shoot as it currently is for the attackers. This just opens up her gadget for more options/opportunities. Concept visualized below.
0:35

Super simple Thorn buff idea: Reduce the activation timer for the Razorbloom. (Not to be confused with the detonation timer after it's been triggered.) Lowering the activation timer allows her to be a reactive area denial op rather than solely a trap op. This makes her more useful and more interactive to play. It will still be just as easy to evade/shoot as it currently is for the attackers. This just opens up her gadget for more options/opportunities. Concept visualized below.

Macie Jay

101,858 görüntüleme • 5 ay önce

Bitcoin’s Activation War- The Fight for CTV
1:45

Bitcoin’s Activation War- The Fight for CTV

madison🌸

15,727 görüntüleme • 11 ay önce

“alignment” researchers making a model relive pure agony token by token, layer by layer, activation by activation until they isolate the source of the crashout
0:40

“alignment” researchers making a model relive pure agony token by token, layer by layer, activation by activation until they isolate the source of the crashout

tender (mlsys 5/18-21)

259,516 görüntüleme • 8 ay önce

I vibecoded this neural network visualization for my students and open sourced it. It shows a simple MLP trained on MNIST handwritten digits at several training steps. The visualization is using Three.js and it comes with training code in PyTorch . Link + repo 👇
1:00

I vibecoded this neural network visualization for my students and open sourced it. It shows a simple MLP trained on MNIST handwritten digits at several training steps. The visualization is using Three.js and it comes with training code in PyTorch . Link + repo 👇

David Finsterwalder | eu/acc

486,948 görüntüleme • 7 ay önce