Optimization

Notes on Gradient Descent

Working through the math of gradient descent — how the loss surface shapes training, and why small implementation details matter more than I expected.