numpy-ML

All algorithms are implemented from scratch using only NumPy, making the code legible and ideal for understanding ML fundamentals, as emphasized in the project description and README.

Covers a wide range from neural networks and reinforcement learning to linear models and preprocessing, providing a one-stop resource for learning diverse ML techniques, as listed in the available models section.

Relies solely on NumPy and the Python standard library, reducing setup complexity and avoiding black-box dependencies, which is perfect for educational environments and prototyping, as stated in the installation notes.

Offers layers, optimizers, losses, and other components that can be combined to create custom neural network architectures, facilitating experimentation, as detailed in the neural networks section.