AI Foundations
Parameter
A value inside a model that is learned during training and influences its behavior.
Example
A language model contains billions of adjustable values learned during training.
Why it matters
It gives AI adjustable values that help it capture many different patterns during learning.
You'll hear it
“This version has more parameters, but we still need to compare the results.”
Can you control it?
Not directly
No direct control. Parameters are values a model learns during training. The people building the model shape them through its design and training data; someone using a finished product does not set them.
Common questions
- Are parameters stored as whole sentences?
- No. Individual values are numbers; meaning is spread across many of them.
- Are parameters the same as settings?
- No. Parameters are learned during training, while settings such as temperature may be chosen when the model is used.
- Do more parameters make a model smarter?
- No. The teaching material, design and task also affect how useful the system is.