suite

Definition

A bundled collection of tasks or evaluations rather than a single test. The rationale is coverage: a method tuned to one task can look strong while failing on adjacent skills, so a suite spreads the measurement across dimensions and reports per-task as well as aggregate scores. Where a benchmark is usually one task with one protocol, a suite is several of them grouped under a shared harness.

Explainers using this term

• Paper Explained: VBVR-Pro — Making "Thinking in Pictures" Trainable, Scorable, and ComparableVBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning