Softmax Function

What:

A function that takes a bunch of numbers and returns them so they all add up to $1$, with each assigned a Probability ~relative to the size of it in the beginning.

The Formula:
$\sigma (x_i)=\frac{e^{x_i}}{{\sum }_{j=1}^n{e}^{x_j}}$

Temperature (For Use in Large Language Models (LLMs))

By dividing the exponents by $T$, it affects how we choose the words - the lower the more deterministic, the higher, the more random.