Neuromancer

👀 Visualize Neural Networks Blazingly Fast in 3D

How it works

• I run a neural network in a seperate webworker using ONNX Web Runtime.
• The neural net has been modified to expose its internal tensors. Output from the neural net is transferred over to main thread using efficient Transferable ArrayBuffer.
• The 3D neural network is updated asynchronously @ 30 FPS.

This dual-thread architecture is also reflected in source:

./web
├── onnx/
├── ui/
├── bridge.ts
└── main.ts

Models You Can Visualize

1. Convolutional neural network for identifying handwritten digits (MNIST dataset)
2. AlexNet (planned)

Models have been modified using ONNX python package.

Note

Work in progress- More models and features are on the way 🫡

Visit the Live Demo or Run Locally

  git clone https://github.com/rwdr0/neuromancer &&\
  cd neuromancer &&\
  npm install &&\
  npm run preview

Inspired by

• These YouTube Videos
• Netron
• Transformer Explainer
• ONNX Runtime Web Demo

Icons attribution: SVGRepo, Wikimedia Commons