🧠
Neural Network Playground
Build and train small neural networks on toy 2D datasets. Watch decision boundaries evolve in real-time.
Neural Network Playgroundalpha
0 parametersArchitecture
Hidden LayersLayer 14
Layer 24
Network
Dataset
Noise0.10
Train ratio80%
Training
Learning rate0.0300
Batch size
Epoch: 0
Class 0Class 1Test
Loss
0.0000
Train Acc
0.0%
Test Acc
0.0%
Epoch
0
Inspired by TensorFlow Playground. Neural network implemented from scratch — no ML libraries. All training runs entirely in your browser.