Margin
Standalone playgrounds

Simulations.

Standalone interactive playgrounds — each focused on one ML idea. They’re separate from the 16-session journey: no progression, no challenges, just open-ended exploration.

Phase 1
1

Teachable Machine

Train an image classifier with your webcam in 30 seconds.

Inspired by Google's Teachable Machine
Phase 2
2

Polynomial Playground

Drag points, watch a curve fit them — see overfitting happen live.

Built for Margin. Inspired by classic bias-variance demos.
Phase 3
3

Neural Net Playground

Build and train a neural network in your browser. No setup.

Inspired by Smilkov & Carter's TensorFlow Playground
Phase 4
4

CNN Explorer

See what each layer of a vision model is looking at.

Inspired by Wang et al.'s CNN Explainer (Polo Club, GA Tech)
Phase 5
5

Transformer Explorer

Type a sentence. Watch attention flow between words in real time.

Inspired by Cho et al.'s Transformer Explainer (Polo Club, GA Tech)
Phase 5
5

Diffusion Explorer

Add noise to an image, then watch the model un-noise it step by step.

Inspired by Distill's diffusion explainersComing soon