minigrad v0.1.0

This is the initial release of minigrad, a toy autograd engine. This project provides a simple minimal framework, for understanding the internals of deep learning frameworks like PyTorch.

Key Features:

• Support for basic tensor operations with automatic differentiation.
• CPU and GPU support using NumPy and CuPy, respectively.
• Simple neural network layers (e.g., Linear) and activations (e.g., ReLU).
• Basic optimizers, including SGD.
• Loss functions, such as MSELoss.

This is meant for learning purposes and is not optimized for speed.