Open Models Almanac
Methodsevaluationevaluation harness

single mention · evaluation

Savanna Benchmark abstraction

Encapsulates benchmark evaluation backends together with result storage and lineage.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See evaluation harness for the methods filed beside it.

Also called Benchmark abstraction.

Filed under evaluation › evaluation harness