Maths for ML / Notation

x \sim p(x)

is distributed as / sampled from

Probability and statisticsBeginner

Say it “x is drawn from p”

x is a random value drawn from the distribution p.

Example

\epsilon \sim \mathcal{N}(0, 1)

In code

rng.normal(0, 1)

torch.randn(())

Where you’ll see it in ML

Noise in diffusion models, weight init, dropout masks, sampling tokens.

Don’t confuse it with

$\approx$ approximately equal Close enough for the purpose at hand, but not exactly equal.
$\mathcal{N}(\mu, \sigma^2)$ normal (Gaussian) distribution The bell curve, set by a mean μ and a variance σ².

← All 94 symbols