https://github.com/ameobea/rnn-viz

Interactive visualizations and demos that are used in a blog post I wrote about logic in the context of neural networks
https://github.com/ameobea/rnn-viz

activation-functions canvas echarts neural-network svelte sveltejs sveltekit tensorflow tf tfjs threejs tinygrad uplot webassembly

Last synced: about 1 month ago
JSON representation

Interactive visualizations and demos that are used in a blog post I wrote about logic in the context of neural networks

• Host: GitHub
• URL: https://github.com/ameobea/rnn-viz
• Owner: Ameobea
• Created: 2022-06-13T03:17:53.000Z (almost 3 years ago)
• Default Branch: main
• Last Pushed: 2024-10-06T05:51:38.000Z (7 months ago)
• Last Synced: 2025-02-27T21:20:04.159Z (about 2 months ago)
• Topics: activation-functions, canvas, echarts, neural-network, svelte, sveltejs, sveltekit, tensorflow, tf, tfjs, threejs, tinygrad, uplot, webassembly
• Language: Jupyter Notebook
• Homepage: https://cprimozic.net/blog/boolean-logic-with-neural-networks/
• Size: 1.82 MB
• Stars: 17
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Neural Network Experiements + Visualizations

![A 5-second looping video showing a spinning 3D wireframe cube.  The cube's corners are labeled with truth tables in the form TTF, TFT, FFF, etc.  The cube is partially filled with solid voxels representing areas where the artificial neuron it's visualizing outputs a value greater than 0.](./static/logic-cube-header.webp)

This repository contains visualizations and demos that are used in a [blog post](https://cprimozic.net/blog/boolean-logic-with-neural-networks/) I wrote about logic in the context of neural networks. The best place to head to get more info on the stuff here is to read that!

It also includes TF.JS implementations of the Ameo activation function as introduced in that post.

Tech used to build these demos includes [uPlot](https://github.com/leeoniya/uPlot) and [eCharts](https://echarts.apache.org/) for charts + graphs, [Three.JS](https://threejs.org/) for the 3D voxel-based function plot seen above, and [TensorFlow.JS](https://github.com/tensorflow/tfjs) for implementing the Ameo activation function and training the binary additional model referenced in the last part of the blog post.