thinQmaster Virtual testing partner Adaptive practice exams that find your weak spots and drill them. Every answer explained. From $69
Enroll

thinQmaster

Find your exam

Five questions free, before you decide.

Step 1 · choose a vendor

In one of our camps? Your practice exams come with the tuition — they are on your study page already.

AI & Machine Learning · Foundational

Model Parameters

The internal values, such as weights and biases, that a model learns from training data.

Definition

Parameters define the function a trained model computes. Large language models are described by parameter count, for example 8 billion or 70 billion parameters, which drives memory and compute needs. Parameters are distinct from hyperparameters, which are set by practitioners.

Example

A 70B model in 16-bit precision needs roughly 140 GB just to hold its weights.

Exam tip

Memory estimate rule of thumb: parameters x bytes per parameter (2 bytes for FP16/BF16).

Related terms