MNIST digit classifier
Draw a handwritten digit and watch a convolutional neural network make its prediction. Explore the probabilities, feature maps, neurons, and learned connections.
- TEST ACCURACY
- See project
- INFERENCE
- In your browser
THE PROJECT COLLECTION
A growing collection of experiments in code, machine learning, and the web. Open a project and explore how it works.
Draw a handwritten digit and watch a convolutional neural network make its prediction. Explore the probabilities, feature maps, neurons, and learned connections.
Watch an array find its order. Explore 15 algorithms, adjust the speed, and follow each operation in the live pseudocode.
Watch simple rules create living patterns. Draw cells, explore classic formations, and let a responsive grid evolve at your pace.
Plant a forest. Start a fire. Watch new growth return. Explore the balance between growth, lightning, and the spread of flames.
Place a start, a flag, and obstacles. Watch six search algorithms explore the grid, with live pseudocode and a path traced step by step.
All five projects run in your browser. No account needed.