AI Glossary Term
What does “Parameter” mean?
Definition
A learned weight inside a model that influences its outputs.
How Parameter works
Example
A model described as '7B' has about seven billion learned numbers. During training each of them is nudged many times, and together they determine how the model responds to any input.
Why it matters
Parameter count is a rough measure of a model's size and cost, though it does not alone determine quality.
Common questions
Parameters vs hyperparameters?
Parameters are learned from data. Hyperparameters are settings you choose before training.
Does more always mean better?
Not always. Data quality, training method, and design also matter, and smaller models can outperform larger older ones.
Where are they stored?
In files called weights or checkpoints, which can be many gigabytes for large models.