← Concept IndexDEFINITION WHY IT MATTERS COMMONLY CONFUSED WITH
Parameters and weights
Also called: model size, billions of parameters
Parameters (also called weights) are the individual numbers a model adjusts during training. 'A 70-billion-parameter model' means it has that many such numbers.
Parameter count is a rough proxy for capability and cost, but not a guarantee of quality — training data, method, and tuning matter as much as raw size.
The context window. Parameters are fixed after training; the context window is the live input space at the moment you use the model.