🧪 Neural Network Training Simulation: Optimization Dynamics

This simulation allows users to observe the complex dynamics of training a neural network by adjusting key parameters such as learning rate, hidden layer size, and neuron count. Witness firsthand how these adjustments impact the optimization process, as reflected in real-time changes in loss and accuracy metrics. The Batch Size slider now drives real mini-batch gradient descent, training on a random subset of the data each epoch instead of always using the full dataset.

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install