Parameter
파라미터(매개변수)
The numbers inside an AI model that are set by training. They usually indicate the size of the model.
Overview
A model is made of a huge number of numbers, and training is the process of tuning them. Each one of those numbers is a parameter. Large models have hundreds of billions of parameters or more.
More parameters usually means the model can do more complex things, but it is also slower and more expensive. Techniques for making small but smart models have improved, so size alone no longer tells you how good a model is.
An everyday example
Imagine a mixing board with hundreds of billions of knobs, and training sets each knob one by one.
Why it matters
Model names often carry a size, like "7-billion-parameter model." It is a way to judge whether a model can run on your own computer or phone.
Also calledweight