Derivatives

Forget all the complex equations from high school. For ML, you only need to know one thing: A derivative is just a slope.

The Ski Slope Analogy

Imagine you are standing on a snowy mountain. The derivative tells you exactly how steep the ground is right under your feet, and which direction points downhill.

Python Implementation

Instead of doing calculus by hand, modern ML uses Automatic Differentiation. Frameworks like [PyTorch](../Ch-9 Deep-Learning-Frameworks/PyTorch.mdx) can magically calculate the exact slope for you!

import torch

# Our location on the mountain (x = 3.0)
# requires_grad=True tells PyTorch to track the slopes!
x = torch.tensor(3.0, requires_grad=True)

# Our mountain shape equation: y = x^2
y = x ** 2

# Calculate the derivative! (dy/dx)
y.backward()

# The slope at x=3 is 6.0!
print("The slope (derivative) is:", x.grad)