[P] A small MLP from scratch in NumPy with a GUI to look inside it while it trains (weight distributions, t-SNE per layer, neuron ablation...) [P]

A developer created an educational tool: a small MLP built from scratch in NumPy. This tool includes a GUI that allows users to visualize the internal workings of the MLP during training, such as weight distributions, t-SNE per layer, and neuron ablation. It uses manual backpropagation, SGD with momentum, L2 regularization, dropout, and cosine decay, achieving approximately 98.5% accuracy on the MNIST dataset with the full training set. The project is available on GitHub.

Unlike many high-level frameworks, this tool offers a unique, from-scratch NumPy implementation with a GUI to visualize internal MLP mechanics during training.