Skip to content

benchmark

CORE

A composite score aggregating a fixed set of language-model evaluation tasks, commonly used to compare small base models.

Current clusters