Term
模型可解释性
Next Token / Wiki

Model Interpretability

The ability and field of study concerned with understanding a model’s behavior, mechanisms or basis for predictions.

Also known as

可解释性

Definition

The ability and field of study concerned with understanding a model’s behavior, mechanisms or basis for predictions.

Usage and distinctions

Changing parameters or improving a test score does not by itself explain all of a model’s internal mechanisms.

Sources